A MAGDM approach based on dual hesitant Q-rung orthopair fuzzy Dombi norm with Hamy mean operators and its application Authors: Xiuqin Ma, Mengyao Wang, Hongwu Qin, Qi Wei, Tao Li Published: 2025 ABSTRACT AbstractAs complex real-world decision-making intensifies, handling inherent fuzziness and uncertainty becomes crucial. Dual hesitant q-rung orthopair fuzzy (DHq-ROF) sets offer a solution for addressing ambiguity and hesitancy for multi-attribute group decision-making (MAGDM). However, the existing MAGDM methods fail to simultaneously consider attribute correlation and flexibility, and involve lower variation rate of the scored values. Additionally, these techniques concern distance measurement with normalization, which leads to information redundancy. To address these above issues, we propose a new MAGDM approach based on DHq-ROF Dombi norm with Hamy Mean (HM) Operators. First, the HM operator considering attribute correlation and the Dombi t-norms and t-conorms (Dt-N&t-CNs) which offer enhanced flexibility are combined within the DHq-ROF environment, proposing DHq-ROF weighted Dombi HM and DHq-ROF weighted Dombi dual HM aggregation operators to obtain comprehensive evaluative information. Second, a new distance measurement method without normalization is put forward to avoid information redundancy, and subsequently, the related entropy measures for objectively deriving the undetermined weights of attributes are presented. Third, this study proposes a novel DHq-ROF-based approach for MAGDM. Finally, a case study on traditional Chinese medicine-based sleep disorder diagnosis confirms the method’s effectiveness, achieving a higher variation rate of the scored values while simultaneously considering flexibility and attribute correlation.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-30997-0.Keywords: Dual hesitant q-rung orthopair fuzzy set, Dombi t-norms and t-conorms, Hamy mean operators, Multi-attribute group decision-makingSubject terms: Engineering, Health care, Mathematics and computing Introduction Research background of TCM diagnosis for SD Sleep Disorders (SD) refers to abnormal amount of sleep as well as the manifestation of abnormal behaviors during sleep, and also the manifestation of disturbances in the normal rhythmic alternation between sleep and wakefulness. SD has become a public health issue of world concern. Statistics from the “2024 White Paper on Chinese Residents’ Sleep Health” released in March 2024 show that the overall sleep quality among the population is poor, and sleep problems are prominent, with the average length of sleep of Chinese people at 6.75 h, and as many as 59% of the population suffering from insomnia, while only 19% of the population having no SD at all1. The causes of SD are mainly biological, such as changes in biological rhythms, physical illnesses, genetic factors, changes in body temperature, etc. Secondly, it is psychological causes, sleep expectation, bad sleep habits and wrong cognition. Thirdly, it is related to stress and changes in the environment, encountering things that are not too favorable in life. SD is very unfavorable to our physical and mental health. Chronic sleep deficiency is associated with elevated risks of developing various health complications, including cardiovascular disorders, metabolic dysfunctions, and mental health conditions. SD has become one of the most prevalent and pressing medical problems facing modern society2–5. TCM is one of the best way to diagnose SD. Traditional Chinese medicine (TCM) diagnoses SDs in various ways, and different methods can be chosen for treatment according to the specific conditions of patients and different causes of SDs. TCM constitutes a holistic medical system encompassing the study of human physiological and pathological mechanisms, while emphasizing disease prevention, clinical diagnosis, therapeutic intervention, and health maintenance6. TCM treatments include medication, acupuncture, massage and tuina, qigong, dietary therapy, and other modalities, which aim to adjust the yin-yang equilibrium in the body, clear the meridians and channels, and harmonize qi and blood, so as to achieve the goal of curing diseases. The 72nd World Health Assembly approved the ICD-11 revision, marking a historic inclusion of Chinese traditional medicine practices in WHO’s global classification system7. Additionally, China also actively participates in the work of the International Organization of TCM, the World Federation of Acupuncture and Moxibustion Societies, and other international TCM organizations, demonstrating the significant influence of TCM in the international arena is enormous8. TCM is a unique system of medicine, characterized by its safety, effectiveness, and wide generalization. It has a significant standing in the world’s medical arena due to its special theoretical system and diagnostic and treatment means, and has provided substantial contributions to the health of humanity. TCM collects information about a patient’s condition through four diagnostic means of looking, smelling, questioning and cutting, and then conducts diagnostic analysis according to the patient’s specific situation to determine the cause, nature, and location of the disease. So as to formulate a personalized and comprehensive treatment plan, the determination of TCM diagnostic plan is also subject to the influence of subjectivity and uncertainty. One of the major features of TCM is that it is different from person to person and dialectical treatment. Individuals are very different from each other and many of them have some other symptoms at the same time, and due to the differences in patients’ individuality. Even for the same disease, patients with different constitution may show different conditions, which leads to potential differences in the treatment results. Therefore, fuzzy MAGDM has an important application in the field of TCM9,10. Development of DHq-ROFSs A significant problem in real-world applications involving MAGDM is managing ambiguous and fuzzy data11. Originally introduced by Zadeh12, fuzzy set (FS) theory emerged as a mathematical framework for handling real-world ambiguity and uncertainty phenomena. It involves the degree of membership (MD), in which elements are part of a particular set with a specific degree of membership. Subsequently, Atanassov and Atanassov13 introduced intuitionistic FSs, which incorporate both MD and non-membership degree (NMD) to represent uncertain information. This approach offers more options for describing phenomena while adhering to the constraint that MD + NMD ≤ 1. In recent years, Yager14,15 expanded upon the classical FS and intuitionistic FS and then proposed Pythagorean fuzzy sets (PFSs) which can represent vagueness and uncertainty across a broader range. However, PFSs also have limitations regarding universality. Consequently, Yager16 introduced q-rung orthopair fuzzy sets (q-ROFSs) by incorporating q-rung relations. It needs to satisfy the constraint that , here, denotes the MD, while represents the NMD. However, due to the more complex problems encountered in the real world, decision-makers (DMs) often hesitate when evaluating alternatives among several possible fuzzy numbers that cannot be represented by a single exact value. In response to this challenge, Torra17 introduced the idea of the hesitant fuzzy set (HFS), a recent expansion of FSs that accommodates multiple values for the degree of MD of an element. Subsequently, recognizing that HFS is limited by its focus solely on MD, Zhu et al.18 introduced the dual hesitant fuzzy set (DHFS). This innovative framework combines the characteristics of IFS and HFS, allowing each element to possess multiple possible values for both MD and NMD, with the requirement that the largest MD plus the largest NMD is no more than 1. With the increasing prevalence of hesitancy and uncertainty in the decision-making process, Xu et al.19 introduced the dual hesitant q-rung orthopair fuzzy set (DHq-ROFS) by extending DHFS based on q-ROFS, which allows each element to possess multiple possible MDs and NMDs, and the sum of the q of the largest MD and the q of the largest NMD is not more than one. When q = 1, DHq-ROFS simplifies to DHFS, and when q = 2, it simplifies to the dual hesitant Pythagorean fuzzy set (DHPFS). The DHq-ROFS model can express the DM’s hesitancy and uncertainty more comprehensively and accurately, offering significant advantages in terms of universality. In practical applications, it assists DMs in addressing complex problems more holistically, thereby greatly enhancing the quality and efficiency of decision-making. It serves as an effective tool for tackling intricate decision-making challenges. The properties of FS, IFS, PFS, q-ROFS, HFS, DHFS, DHPFS, and DHq-ROFS are compared in Table 1. This comparison clearly demonstrates that DHq-ROFS possesses superior properties for managing complex decision-making problems. Table 1Comparing the DHq-ROFS model to other current fuzzy set models.ModelVaguenessHesitationFalsityNumbers of MD and NMD are more than oneCapturing information by q-ROFUniversalityFS√IFS√√PFS√√q-ROFS√√√√HFS√√DHFS√√√√DHPFS√√√√DHq-ROFS√√√√√√ Comparing the DHq-ROFS model to other current fuzzy set models. A brief review of MAGDM One of the most successful applications of DHq-ROFS, involves MAGDM, which effectively addresses complex problems20–23. MAGDM is a decision issue of identifying the most desirable option or ordering candidate solutions by simultaneously considering the opinions of different DMs simultaneously with the consideration of multiple attributes, which is an important decision-making method in decision science. As evidenced by prior research, MAGDM has been extensively utilized in domains. For example, Fu and Xiao24achieved optimal selection of projects such as venture capital by building a MAGDM in an interval-valued intuitionistic fuzzy context. Arya and Kumar25 MAGDM to handle the supplier selection issue in a q-rung orthopair fuzzy setting. He et al.26 employed the MAGDM approach to tackle the problem of financial management performance evaluation for the purpose of identifying investment opportunities. Tian et al.27 presented a MAGDM approach relying on flexible linguistic representation and used it for the assessment of urban flood risks. By analyzing specific cases, we can gain insights into the effects and challenges of MAGDM methods in practical applications. The importance of MAGDM is that it enables DMs to systematically handle complex problems, thus improving the efficiency and precision of decision-making. A brief description of Dt-N&t-CNs and HM aggregation AOs The Aggregation Operator (AO) function, which combines and integrates multiple data items to produce a single output value during the data fusion process, plays a crucial role in MAGDM. An excellent AO known as Dt-N&t-CNs was introduced by Dombi28. First, the Dombi parameter can be adjusted according to the needs of DM, which enhances the flexibility of the information aggregation process. Secondly, Dt-N&t-CNs efficiently handles complex and uncertain data, providing a robust and reliable solution for data fusion. Numerous researchers have conducted in-depth studies on Dt-N&t-CNs across various fuzzy domains. Akram et al.29 proposed the Pythagorean Dombi fuzzy weighted arithmetic averaging operator, the Pythagorean Dombi fuzzy weighted geometric averaging operator, and four additional operators, addressing the Multi-Criteria Decision-Making problem under PFS. Jana et al.30 described the q-ROF Dombi-weighted averaging, order-weighted averaging, and hybrid weighted averaging, along with their geometric variants in the q-ROF context. Subsequently, Jana et al.31 introduced Pythagorean fuzzy power Dombi weighted averaging, order-weighted averaging, and hybrid weighted averaging operators, as well as their geometric variants. Seikh and Mandal have contributed to fuzzy decision-making by developing Dombi operations for intuitionistic and Fermatean sets32,33, Frank operators for q-ROF sets34, and by applying these within methods like EDAS to solve selection problems35–38. In actual MAGDM problems, most of the attributes being fused are strongly correlated with each other. In 1890, Hamy39 proposed the Hamy symmetric function, which was later extended to the r-order HM; in 1998, Hara et al.40 improved the inequality of arithmetic mean and geometric mean, proving that HM and its dual value are monotonically between arithmetic mean and geometric mean. The parameters of the HM operator can be adjusted according to the specific needs of decision-making, which enhances its flexibility and yields results that are more aligned with the actual situation. Compared to the Bonferroni Mean and the Heronian Mean, the HM offers distinct advantages. While the Bonferroni Mean and the Heronian Mean primarily focus on the interrelationships between pairs of attributes, HM takes into account the interrelationships among multiple attributes. This characteristic makes HM highly flexible and generalized, rendering it more suitable for complex decision-making problems in practical applications. As a result, it has garnered significant attention from scholars. Xing et al.41 combined the HM operator with interaction operational laws to construct a q-ROF operator, which are applied to solve MAGDM problems. Garg et al.42 developed a series of complex Pythagorean fuzzy aggregation operators, encompassing both basic and dual forms with weighted variants under the HM framework. Hussain et al.43 presented a series of IF Aczel-Alsina HM operators by integrating the HM operator with the Aczel-Alsina norm in an IF environment. Recent works by Ali et al. have advanced the development of aggregation operators, introducing Hamy mean, copula-based power, and partitioned Maclaurin symmetric mean operators for various q-rung orthopair and other fuzzy set extensions44–48. The motivation of this study In summary, the primary research motivations underlying this work can be articulated as follows: After a review based on the existing literatures, it is clear that there still are some gaps in the existing research. First, diagnosing SD in TCM faces two primary challenges: ambiguous symptoms and physician uncertainty. This issue arises primarily because patients exhibit a wide range of symptoms and often struggle to articulate their concerns clearly. Specifically, patients frequently find it difficult to accurately enumerate their symptoms or distinguish between similar underlying causes of SD, while physicians may feel hesitant when making diagnoses based on these vague descriptions. Consequently, final diagnoses often become compromises made amid significant uncertainty. Although fuzzy set theory has been utilized in medical diagnosis previously49–51, current methodologies still fail to adequately address TCM’s dual challenges: unclear symptom descriptions and uncertain diagnostic decisions. First, diagnosing SD in TCM faces two primary challenges: ambiguous symptoms and physician uncertainty. This issue arises primarily because patients exhibit a wide range of symptoms and often struggle to articulate their concerns clearly. Specifically, patients frequently find it difficult to accurately enumerate their symptoms or distinguish between similar underlying causes of SD, while physicians may feel hesitant when making diagnoses based on these vague descriptions. Consequently, final diagnoses often become compromises made amid significant uncertainty. Although fuzzy set theory has been utilized in medical diagnosis previously49–51, current methodologies still fail to adequately address TCM’s dual challenges: unclear symptom descriptions and uncertain diagnostic decisions. While significant progress has been made in developing decision-making operators for DHq-ROFSs, existing methodologies52–58 exhibit critical limitations in addressing the interrelationships among attributes that are prevalent in complex situations. Current operators, however, rely on overly simplistic independence assumptions that fail to accurately model these dynamic interrelationships. Furthermore, despite the theoretical potential of Dt-N&t-CNs frameworks combined with HM operators for capturing attribute correlations, no structured integration has been achieved. This deficiency emphasizes the immediate requirement to develop innovative AOs that not only account for the interrelationships among fusion parameters but also incorporate the advantages of flexibility, thereby enhancing stronger distinguishing power for alternatives ranking. While significant progress has been made in developing decision-making operators for DHq-ROFSs, existing methodologies52–58 exhibit critical limitations in addressing the interrelationships among attributes that are prevalent in complex situations. Current operators, however, rely on overly simplistic independence assumptions that fail to accurately model these dynamic interrelationships. Furthermore, despite the theoretical potential of Dt-N&t-CNs frameworks combined with HM operators for capturing attribute correlations, no structured integration has been achieved. This deficiency emphasizes the immediate requirement to develop innovative AOs that not only account for the interrelationships among fusion parameters but also incorporate the advantages of flexibility, thereby enhancing stronger distinguishing power for alternatives ranking. Significant limitations remain in current decision-making methodologies concerning distance measurement and the determination of attribute weights. Although most studies59,60 necessitate data normalization to achieve dimensional homogeneity, this preprocessing step inevitably leads to information distortion and feature redundancy. Simultaneously, the common practice of directly assigning attribute weights52–54,56,57 introduces considerable subjective bias. Significant limitations remain in current decision-making methodologies concerning distance measurement and the determination of attribute weights. Although most studies59,60 necessitate data normalization to achieve dimensional homogeneity, this preprocessing step inevitably leads to information distortion and feature redundancy. Simultaneously, the common practice of directly assigning attribute weights52–54,56,57 introduces considerable subjective bias. While the MAGDM methodology serves as a fundamental approach for addressing decision-making processes, its practical application in complex decision-making environments requires greater flexibility to accommodate the diverse risk preferences of DMs. Therefore, identifying an efficient MAGDM method is essential. While the MAGDM methodology serves as a fundamental approach for addressing decision-making processes, its practical application in complex decision-making environments requires greater flexibility to accommodate the diverse risk preferences of DMs. Therefore, identifying an efficient MAGDM method is essential. Based on the analysis presented above, this research makes the following main contributions and innovations: The main contributions and innovations of this study We employ DHq-ROFSs as a comprehensive data characterization framework. This framework effectively captures the ambiguity and hesitancy inherent in the data, making it particularly suitable for TCM diagnosis of SD. The dual characterization capability enables the simultaneous representation of patients’ symptom ambiguity and DMs’ diagnostic hesitancy, thereby providing the most appropriate data model for TCM diagnosis of SD. We employ DHq-ROFSs as a comprehensive data characterization framework. This framework effectively captures the ambiguity and hesitancy inherent in the data, making it particularly suitable for TCM diagnosis of SD. The dual characterization capability enables the simultaneous representation of patients’ symptom ambiguity and DMs’ diagnostic hesitancy, thereby providing the most appropriate data model for TCM diagnosis of SD. To address the interrelationships among multiple parameters in decision-making, this study develops two novel AOs: the DHq-ROF weighted Dombi HM (DHq-ROFWDHM) and the DHq-ROF weighted Dombi dual HM (DHq-ROFWDDHM). These operators leverage the inherent capacity of the HM to model multiple parameter correlations, along with the operational flexibility of the Dt-N&t-CNs through adjustable Dombi parameters. Therefore, the proposed two operators not only account for the interrelationships among fusion parameters but also incorporate the advantages of flexibility. Moreover, they improve the variation rate of the scored values, thereby offering stronger distinguishing power for alternative ranking. To address the interrelationships among multiple parameters in decision-making, this study develops two novel AOs: the DHq-ROF weighted Dombi HM (DHq-ROFWDHM) and the DHq-ROF weighted Dombi dual HM (DHq-ROFWDDHM). These operators leverage the inherent capacity of the HM to model multiple parameter correlations, along with the operational flexibility of the Dt-N&t-CNs through adjustable Dombi parameters. Therefore, the proposed two operators not only account for the interrelationships among fusion parameters but also incorporate the advantages of flexibility. Moreover, they improve the variation rate of the scored values, thereby offering stronger distinguishing power for alternative ranking. The third contribution of this study lies in its work conducted within the DHq-ROFS environment, which features: (1) a distance metric that preserves the original data distributions without normalization constraints, and (2) entropy weighting mechanisms that objectively derive weights through information entropy analysis. This innovation ensures both metric integrity and decision objectivity in SD diagnosis. The third contribution of this study lies in its work conducted within the DHq-ROFS environment, which features: (1) a distance metric that preserves the original data distributions without normalization constraints, and (2) entropy weighting mechanisms that objectively derive weights through information entropy analysis. This innovation ensures both metric integrity and decision objectivity in SD diagnosis. A novel MAGDM model is proposed. Through comprehensive sensitivity examinations and comparative evaluations with state-of-the-art methods, the approach demonstrates significant performance advantages when addressing complex decision-making challenges. A TCM diagnosis case study further confirms its clinical applicability, illustrating the model’s ability to generate personalized treatment plans by integrating ambiguous symptoms with clinical expertise. This approach effectively links theoretical frameworks with practical diagnostic requirements. A novel MAGDM model is proposed. Through comprehensive sensitivity examinations and comparative evaluations with state-of-the-art methods, the approach demonstrates significant performance advantages when addressing complex decision-making challenges. A TCM diagnosis case study further confirms its clinical applicability, illustrating the model’s ability to generate personalized treatment plans by integrating ambiguous symptoms with clinical expertise. This approach effectively links theoretical frameworks with practical diagnostic requirements. Organization of the study The paper is organized in the following manner: “Preliminaries” provides the necessary definitions and examines core properties of fundamental notions. “Dombi HM operators in DHq-ROF frameworks” presents the DHq-ROFWDHM and DHq-ROFWDDHM aggregation operators within the DHq-ROFS framework and examines their properties. “New distance formula and entropy-based method for determining attribute weights” introduces a novel distance formula and a method for determining unknown attribute weights based on entropy. “Developments of methodologies MAGDM in DHq-ROF frameworks” outlines a new MAGDM framework. In “Practical application”, through the analysis of a case study on the TCM diagnosis of SD, the effectiveness and applicability of the proposed strategy in dealing with practical MAGDM challenges are demonstrated. “Sensitivity analysis” validates the method’s applicability through sensitivity analysis. “Comparative analysis” conducts a comparative analysis, demonstrating that the newly proposed MAGDM method exhibits significant advantages in terms of characteristics, superiority, and flexibility compared to many state-of-the-art methods in practical applications. Finally, “Conclusions and future research directions” discusses the conclusions and outlines future research directions. Preliminaries This segment provides a brief discussion on several fundamental concepts relevant to the topic. Based on q-ROFSs as proposed by Yager16 and DHFSs introduced by Zhu et al.18, the theoretical framework of DHq-ROFSs was introduced by Xu et al.19, along with the formulation of essential operational rules. Definition 2.1 (Xu et al.19). Let denote a universal of discourse. A DHq-ROFS, on is presented as: where and are two sets of real numbers in , They respectively represent the probable MD and NMD of the element with respect to the set , subject to the following conditions: For convenience, pairis referred as a DHq-ROF number (DHq-ROFN)19 and is represented by . Definition 2.2 (Xu et al.19). Let represent a DHq-ROFN. The score function is defined as follows: And the accuracy function of , denoted as , is defined as: where and are the number of in and respectively. The steps to order DHq-ROFNs are presented as follows: Let and be any two DHq-ROFNs, If , then is superior to , indicated by ; If , then is superior to , indicated by ; If , then.When , then ;When , then is equivalent to , denoted by . If , then.When , then ;When , then is equivalent to , denoted by . When , then ; When , then ; When , then is equivalent to , denoted by . When , then is equivalent to , denoted by . The HM and DHM operators In 1998, Hara et al.40 initially proposed the HM operator, which takes into account the interrelationships among various fusion parameters. Definition 2.3 (Hara et al.40). For a collection of crisp numbers , and , The HM operator is expressed as: where represents all the -tuple combinations of , and denotes the binomial coefficient. The parameter specifically controls the number of attributes considered to be interrelated in each calculation. Crucially, the combinatorial term ensures that the aggregation incorporates the synergistic effects from every possible subset of attributes. This intrinsic design makes the HM operator uniquely powerful for modeling the complex, higher-order interrelationships that exist among symptoms in TCM diagnosis. The HM operator has four properties. , if for all i. , if for all i. The following presents two specific cases of the HM operator. When , , which is the arithmetic mean operator. When , , which is the arithmetic mean operator. When , , which is the geometric mean operator. When , , which is the geometric mean operator. Furthermore, Wu et al.61 generalized the HM operator through dual operation principles, introducing the dual Hamy mean (DHM) operator defined as follows. Definition 2.4 (Wu et al.61). For a collection of crisp numbers , and , The DHM operator is expressed as: where represents all the -tuple combinations of and denotes the binomial coefficient. The DHM operator has four properties. , if for all i. , if for all i. The following presents two specific cases of the DHM operator. When , , which is the geometric mean operator. When , , which is the geometric mean operator. When , , which is the arithmetic mean operator. When , , which is the arithmetic mean operator. Dt-N&t-CNs Definition 2.5 (Dombi28). For any real numbers and , the Dombi t-norm and Dombi t-conorm , are defined as: Operations on DHq-ROFNs based on Dt-N&t-CNs According to Dt-N&t-CNs, Sarkar and Biswas62 proposed several mathematical operations under DHq-ROFNs, which are defined as shown below. Definition 2.6 (Sarkar and Biswas62). Suppose , , and represent any three DHq-ROFNs and . Then, Dombi HM operators in DHq-ROF frameworks This section introduces the DHq-ROFWDHM and DHq-ROFWDDHM aggregation operators by combining Dt-N&t-CNs and HM within the MAGDM framework, along with an analysis of their properties. Definition 3.1 Let for be a set of DHq-ROFNs, be the weight vector of , such that and , with . Then the DHq-ROFWDHM operator is defined as where traverse all the -tuple combinations of and denotes the binomial coefficient. We now discuss the characteristics of the newly developed DHq-ROFWDHM operator. Theorem 3.1 Let for be a set of DHq-ROFNs, be the weight vector of , such that and , with . Then the aggregated value obtained by the DHq-ROFWDHM operator is also a DHq-ROFN, and where The demonstration of Theorem 3.1 is provided in the Appendix section. The theorem’s validity is confirmed through Example 3.1. Example 3.1 Consider the set , , which consists of three DHq-ROFNs. Let , and be the weighted vector of these three DHq-ROFNs. Subsequently, we utilize the proposed DHq-ROFWDHM operator to aggregate the elements and in set P as follows: Now, the characteristics of the proposed DHq-ROFWDHM operator are designed in the following way. Theorem 3.2 (Monotonicity). Let for and for be two collections of DHq-ROFNs. If and for all , then . The demonstration of Theorem 3.2 is provided in the Appendix section. Theorem 3.3 (Boundary). Let for be a set of DHq-ROFNs. If and then The demonstration of Theorem 3.3 is provided in the Appendix section. Next, DHM and Dombi operations are combined to create the DHq-ROFWDDHM AO. Definition 3.2 Let for be a set of DHq-ROFNs, be the weight vector of , such that and , with . Then the DHq-ROFWDDHM operator is defined as where traversal all the -tuple combinations of and is the binomial coefficient. We now discuss the characteristics of the newly developed DHq-ROFWDDHM operator. Theorem 3.4 Let for be a set of DHq-ROFNs, be the weight vector of , such that and , with . Then the aggregated value obtained by the DHq-ROFWDDHM operator is also a DHq-ROFN, and The proof of Theorem 3.4 is identical to the proof of Theorem 3.1. Note 3.1 Similar reasoning demonstrates that the DHq-ROFWDDHM operator equally fulfills the monotonicity and boundedness conditions. Applying the DHq-ROFWDDHM operator to Example 3.1 yields the following aggregation result: New distance formula and entropy-based method for determining attribute weights In this part, firstly, an innovative distance formula is proposed for calculating the distance between any two DHq-ROFNs without normalization. Definition 4.1 Let , and represent any two DHq-ROFNs, then the DHq-ROF distance between and is defined as where , represent the counts of hesitant elements in and respectively. while , represent the corresponding counts for and , respectively. Besides, , , , and . It is easy to prove that the distance measure between any two DHq-ROFNs satisfies the following properties: (Non-negativity) . (Non-negativity) . (Identity of indiscernibles) . (Identity of indiscernibles) . (Symmetry) . (Symmetry) . The proof of the properties mentioned above is presented in Appendix. Remark 4.1 The distance measures proposed by Wang et al.60 and Kou et al.59 requires that: compared sets must have equal numbers of MDs and their NMDs must also be equal in quantity. However, in many practical scenarios, the quantities of MDs and NMDs are unequal. This discrepancy necessitates the addition of shorter DHq-ROFNs. A common complementary strategy is to adjust either the maximum or minimum values to attain identical numbers of MDs and NMDs. However, such adjustments may lead to distortion and redundancy in the original data. Therefore, the distance formula given in Definition 4.1 is more reasonable and practical. To enhance understanding of the aforementioned distance formula, Example 4.1 below offers a further demonstration of the operation. Example 4.1 For the DHq-ROFNs pair: , , Applying Definition 4.1 (let q=3), their distance can be calculated as follows: Entropy measurement is an effective method for calculating weights. In the context of MAGDM, each alternative may possess multiple attributes, each exerting varying degrees of influence on the final decision outcome. The weights assigned to these attributes reflect their significance during the MAGDM process. Furthermore, the entropy metric provides a quantitative assessment of the uncertainty inherent in each attribute’s influence on the final decision. By calculating the entropy value for each attribute, it becomes possible to assess their importance in decision-making and subsequently determine the corresponding weights. The entropy measure not only assists decision-makers in quantifying the influence of each attribute on the outcome but also captures the preferences and significance of different attributes to the DM59,63. By judiciously determining these weights, the decision-making results can be rendered more scientific, rational, and equitable. Thus, the current study develops a new entropy measure building upon the introduced DHq-ROFN distance formula, integrated with a complementary attribute weight calculation scheme. The underlying axiom is stated below. Definition 4.2 For any two DHq-ROFNs , a real-valued function qualifies as an entropy measure for DHq-ROFNs if it satisfies the following axiomatic conditions. =0, iff is either or ; =0, iff is either or ; =1, iff and , where and are the th smallest values of and , respectively; =1, iff and , where and are the th smallest values of and , respectively; whenever ; whenever ; Based on this axiom, the entropy of a DHq-ROFN is computed as: where denotes the distance between and its complement Example 4.2 For the DHq-ROFN , then . The entropy is calculated according to Definition 4.2 with parameter q=3. Then, it can be calculated that Based on the entropy measure in Eq. (11), a new method is proposed to determine the weights of aggregated DHq-ROFNs. For a collection of DHq-ROFNs denoted by , the corresponding weights of is computed as follows: Developments of methodologies MAGDM in DHq-ROF frameworks This section presents a novel MAGDM methodology within the DHq-ROF framework, with complete implementation details following. Consider a MAGDM problem with m treatment alternatives denoted by , n evaluation attributes , and p DMs . The attribute importance is characterized by a normalized weight vector satisfing with , while the DMs’ weights are represented by where and . The evaluation information is expressed through DHq-ROF decision matrices (DHq-ROFDM) , where each element consists of the MD set and NMD set , representing the -th DM’s assessment of alternative under attribute . Based on this formulation, the proposed methodology is developed through the following procedural steps: Step 1. Multiple DMs construct multiple DHq-ROFDMs based on various attributes and alternatives. Step 2. For the MAGDM problem, the attributes are classified into cost and benefit-types, and the various types of evaluation matrices are normalized using the following methods: represents the complement of , i.e., . Step 3. If the weight of the attribute is not provided in advance, make use of the entropy method set forth in Sect. 4 to determine the attribute’s weight. Step 4. Utilize DHq-ROFWDHM (or DHq-ROFWDDHM) operator to aggregate the values for each alternative across all the individual DHq-ROFDMs. The aggregation is performed as follows: (or . Step 5. Calculate the scores for the overall DHq-ROFNs . Step 6. Calculate the final score of each alternative by aggregating the DM’ weighted scores. Let be the set of p DMs, with their corresponding weight vector , where and . For each alternative , the final score is computed as the weighted sum of the scores from all DMs: Where denotes the score of alternative evaluated by DM , obtained in Step 5. Example 5.1 For instance, consider two DMs with weights and two alternatives and . Suppose the scores from the two DMs are , for , and , for . Then the final score are calculated as: These final scores are then used for ranking the alternatives in Step 7. Step 7. Rank the all alternatives and select the best one(s) based on the scores . To address the scenario where score values of multiple alternatives are equal, the following supplementary ranking rules are adopted: If , (i.e., the scores of alternatives and are equal), further compare their accuracy functions and (defined in Eq. (3)). If , then alternative is superior to , denoted as ; If , then alternative is superior to , denoted as ; If , and are considered equivalent. In this case, further screening can be conducted by integrating practical TCM clinical scenarios, such as treatment cost, patient constitution adaptability, and long-term therapeutic sustainability. If , and are considered equivalent. In this case, further screening can be conducted by integrating practical TCM clinical scenarios, such as treatment cost, patient constitution adaptability, and long-term therapeutic sustainability. Practical application Description of the problem TCM plays a crucial role in diagnosing SD. TCM employs various diagnostic methods for SD and selects different treatment approaches based on the patient’s specific conditions and the underlying causes of SD. A key characteristic of TCM is its personalized nature. The dialectical treatment provides individualized care tailored to each patient. This holistic perspective and focus on personalized treatment confer unique advantages and effectiveness in TCM’s approach to managing SD. When formulating a treatment plan, TCM practitioners consider the following aspects: In TCM, each disease is believed to have its own unique cause, pathological mechanism, and developmental process. Therefore, it is essential to assess the appropriateness of the treatment method. This involves determining whether the chosen treatment aligns with the disease’s pathological mechanism and whether it can effectively alleviate or eliminate the symptoms64. In TCM, each disease is believed to have its own unique cause, pathological mechanism, and developmental process. Therefore, it is essential to assess the appropriateness of the treatment method. This involves determining whether the chosen treatment aligns with the disease’s pathological mechanism and whether it can effectively alleviate or eliminate the symptoms64. Therapeutic efficacy is the primary criterion for assessing the effectiveness of TCM treatment. It encompasses two main aspects: first, near-term efficacy, which refers to the improvement of symptoms shortly after treatment; and second, long-term efficacy, which pertains to the stabilization of the condition over an extended period (e.g., several months or years) following treatment65. Therapeutic efficacy is the primary criterion for assessing the effectiveness of TCM treatment. It encompasses two main aspects: first, near-term efficacy, which refers to the improvement of symptoms shortly after treatment; and second, long-term efficacy, which pertains to the stabilization of the condition over an extended period (e.g., several months or years) following treatment65. The therapeutic methods used in TCM generally have a high degree of safety. However, variations in patients’ physiques and conditions may result in adverse reactions or side effects during treatment. Therefore, it is essential to monitor potential adverse reactions and side effects, as well as their incidence, throughout the course of TCM treatment66. The therapeutic methods used in TCM generally have a high degree of safety. However, variations in patients’ physiques and conditions may result in adverse reactions or side effects during treatment. Therefore, it is essential to monitor potential adverse reactions and side effects, as well as their incidence, throughout the course of TCM treatment66. Compared to the high treatment costs associated with modern medicine, TCM treatments are relatively affordable. However, the effectiveness of these treatments is directly linked to their cost. Therefore, factors such as treatment cost, efficacy, and patient satisfaction must be considered to evaluate the economic benefits and overall value of different treatment options67. Compared to the high treatment costs associated with modern medicine, TCM treatments are relatively affordable. However, the effectiveness of these treatments is directly linked to their cost. Therefore, factors such as treatment cost, efficacy, and patient satisfaction must be considered to evaluate the economic benefits and overall value of different treatment options67. TCM diagnosis necessitates the consideration of multiple factors. Collectively, these attributes form a comprehensive index system for evaluating treatment options, which aids in the scientific and thorough assessment of the strengths and characteristics of various treatments. This approach enables the development of a more personalized and holistic treatment plan for patients. Selecting an appropriate TCM treatment plan can be challenging due to the diverse patient-specific conditions and underlying causes of syndromes. Additionally, the determination of TCM diagnostic options is influenced by subjectivity and uncertainty. Therefore, it is crucial to utilize the new MAGDM model to effectively tackle this complex issue. Solution to the problem The four alternative treatment options for TCM diagnosis of SD are as follows: Traditional Chinese Medicine (): This approach is grounded in TCM theories and aims to treat the disease by selecting Chinese herbs with specific medicinal properties and efficacy. These herbs are formulated, decocted, or prepared and administered to the patient orally, externally, or through other methods. Acupuncture (): This technique involves stimulating acupuncture points on the human body to adjust the flow of qi and blood, as well as to restore the balance of yin and yang. Massage (): This method stimulates specific areas of the body through manipulation to promote the flow of qi and blood, alleviate muscle tension, and relieve pain and other symptoms. Diet Therapy (): A distinctive treatment method in TCM, Diet Therapy aims to treat diseases and promote health by adjusting dietary habits and food combinations. Cupping Therapy (): This technique primarily targets acupuncture points and meridians on the back to clear blockages and restore the balance of yin and yang in the internal organs. Cupping is particularly effective for addressing issues related to the invasion of wind, cold, and dampness, as well as qi and blood stagnation, which can lead to sleep disorders. Evaluation attributes include: Disease Matching (), Therapeutic efficacy (), Safety (), and Cost-Effectiveness (). Three DMs evaluate a patient and their weights are . For convenience, ; Dombi parameter, ; and HM parameter, . The decision-making procedure is implemented through the following operational steps: Step 1. The three DMs construct three DHq-ROFDMs based on the patient’s specific situation through attributes and alternative treatment alternatives. These results are presented in Tables 2, 3 and 4. Table 2DHq-ROFDM of . DHq-ROFDM of . Table 3DHq-ROFDM of . DHq-ROFDM of . Table 4DHq-ROFDM of . DHq-ROFDM of . Step 2. Since all five attributes are revenue types, the three DHq-ROFDMs do not need to be normalized. Step 3. Using Eqs. (11)-(12), we compute the entropy weights for each attribute across all alternatives . The complete weight distributions are tabulated in Tables 5, 6 and 7. Table 5The attribute’s weight of . The attribute’s weight of . Table 6The attribute’s weight of . The attribute’s weight of . Table 7The attribute’s weight of . The attribute’s weight of . Step 4. The DHq-ROFWDHM operator aggregates the evaluation values from individual DHq-ROFDMs for each alternative . Due to the sheer volume and complexity of the data, it is omitted here. Step 5. For every alternative within each DHq-ROFDM, we compute the score value , with complete results documented in Table 8. Table 8The score values of evaluation information using DHq-ROFWDHM operator. The score values of evaluation information using DHq-ROFWDHM operator. Step 6. Final scores are calculated using expert weights . Step 7. Based on the computed scores, the alternatives are prioritized as , therefore, the best treatment option for this patient with SD is . Here, it should be pointed out that when the MAGDM problem described above was resolved with the DHq-ROFWDDHM operator, the obtained score values are: 0.1237,0.0872, 0.6351, 0.1558, and 0.2984, and the final ranking result is .Thus, it is again verified that the the best treatment option for this patient with SD is . Sensitivity analysis Sensitivity of decision-making results to HM parameter This study investigates the influence of parameter in the HM operator on decision outcomes through the proposed methodology. Tables 9 and 10 present the score results of varying values while keeping and constant. This analysis uses both DHq-ROFWDHM and DHq-ROFWDDHM operator to respectively get different scoring values. Table 9Score results obtained by the DHq-ROFWDHM operator as changes (q = 3, t = 2).Varying Ranking order Score results obtained by the DHq-ROFWDHM operator as changes (q = 3, t = 2). Table 10Score results obtained by the DHq-ROFWDDHM operator as changes (q = 3, t = 2).Varying Ranking order Score results obtained by the DHq-ROFWDDHM operator as changes (q = 3, t = 2). The HM parameter was selected from the range29,46 to systematically examine how varying levels of attribute interrelationships affect decision outcomes. This parameter range allows us to demonstrate how the proposed operators transition between different aggregation behaviors while maintaining the capability to capture correlations among multiple attributes simultaneously. When the DHq-ROFWDHM operator is applied, the rank results of different alternatives for TCM diagnosis of sleep disorders are presented in Fig. 1. It is evident that the rankings of , , and experience slight changes with the increase of parameter . However, these changes do not impact the overall decision-making results. For all values of within the range29,46, the ranking results consistently show that the optimal solution remains superior to the worst solution . Therefore, in this context, is identified as the best treatment option. Ranking results obtained by the DHq-ROFWDHM operator as changes (q = 3, t = 2). When the DHq-ROFWDDHM operator is employed, the rank results of every alternative are displayed in Fig. 2. It is evident that for all , the ranking of the optimal solution in relation to the second-best solution remains consistent. Consequently, is the superior treatment option in all scenarios. Ranking results obtained by the DHq-ROFWDDHM operator as changes (q = 3, t = 2). The optimal solution remains invariant to variations in parameter , demonstrating the robustness of both HM and DHM operators. Notably, the Hamy operator captures attribute interrelationships, as evidenced by -dependent ranking variations among intermediate alternatives. This enables decision-makers to align values with their risk tolerance levels. Sensitivity of decision-making results to Dombi parameter t Current research focuses on examining how the Dombi parameter influences decision outcomes when employing DHqROFWDHM and DHqROFWDDHM operators. When considering the conditions and , the score results of the sorting are shown in Tables 11 and 12 respectively. Table 11Score results obtained by the DHq-ROFWDHM operator as changes (q = 3,).Varying t Ranking order Score results obtained by the DHq-ROFWDHM operator as changes (q = 3,). Table 12Score results obtained by the DHq-ROFWDDHM operator as changes (q = 3,).Varying t Ranking order Score results obtained by the DHq-ROFWDDHM operator as changes (q = 3,). The Dombi parameter t was examined over the range43,44 to evaluate operational flexibility while ensuring computational stability. Although Dombi operations remain mathematically valid for all >0, we selected ≥2 following established literature conventions and to avoid potential numerical instabilities that can occur with smaller values. This range sufficiently captures the essential characteristics of Dombi norm flexibility while providing clear insights into parameter sensitivity. First, when the DHq-ROFWDHM operator is applied and the parameter is varied within the range in43,44, the corresponding rank results for the various treatment options are illustrated in Fig. 3. Notably, the ranking of the treatment plans remains consistent (i.e., ) as the value of the Dombi parameter increases. This stability in ranking results indicates that the DHq-ROFWDHM operator produces reliable outcomes under the conditions q=3 and . Ranking results obtained by the DHq-ROFWDHM operator as changes (q = 3,). Then, by applying the DHq-ROFWDDHM operator, Fig. 4 shows the rank results corresponding to different treatment options with the change of the -value. When =2, the figure shows the ranking of alternatives as . The ranking remains stable as for . It is worth noting that is consistently the best alternative in all scenarios. Ranking results obtained by the DHq-ROFWDDHM operator as changes (q = 3,). The analysis reveals that while alternative scores exhibit -value dependency, their relative rankings maintain stability. Decision-makers may select -values according to their preference profiles. Figures 3 and 4 further demonstrate enhanced ranking discrimination at higher -values. Sensitivity of decision-making results to DHq-ROFNs parameter q In this subsection, with fixed parameters (, ), we examine the -parameter’s influence on decision outcomes using DHqROFWDHM and DHqROFWDDHM operators. Tables 13 and 14 respectively display the scoring results under varying -values for both operators. Table 13Score results obtained by the DHq-ROFWDHM operator as changes (t = 2,).Varying q Ranking order Score results obtained by the DHq-ROFWDHM operator as changes (t = 2,). Table 14Score results obtained by the DHq-ROFWDDHM operator as changes (t = 2,).Varying q Ranking order Score results obtained by the DHq-ROFWDDHM operator as changes (t = 2,). Parameter q was tested across28,29 to validate the method’s performance across different uncertainty representation spaces. This range covers the transition from dual hesitant fuzzy sets (q = 1) through Pythagorean fuzzy sets (q = 2) to more general orthopair fuzzy environments, demonstrating the proposed approach’s adaptability to varying levels of information certainty while maintaining the constraint throughout the analysis. When the DHqROFWDHM operator is applied and the value of is varied, Fig. 5 shows the ranking results of different treatment alternatives. As depicted in the figure, when , the ranking is . For and within the range of , the alternatives ranking is , indicating that only the positions of and have changed, while remains the best treatment plan. Ranking results obtained by the DHq-ROFWDHM operator as changes (t = 2,). The ranking results of different treatment alternatives are shown in Fig. 5. As can be seen from this figure. Next, when the DHqROFWDDHM operator is utilized, the ranking results of all the alternatives are presented in Fig. 6. As illustrated in the figure, when in , the ranking of the alternatives fluctuates, with the best alternative being . However, when in , the ranking of the alternatives stabilizes (i.e., ), with remaining the top choice. Ranking results obtained by the DHq-ROFWDDHM operator as changes (t = 2,). Thus, the optimal treatment plan remains unchanged regardless of the value; however, there are slightly different alternatives in between, which demonstrates the stability of the operators. Comparative analysis In this section, a rigorous comparative analysis demonstrates the significant advantages of the newly proposed MAGDM method over five prominent state-of-the-art approaches: those developed by (Gayen et al.; Gayen et al.; Zhou et al.; Wang et al.; Hussain et al.; Wang et al.; Kavitha et al.)52–58. Correctness verification This study conducts a systematic comparison between the proposed MAGDM approach and established methods to verify its validity. Prior works include: Gayen et al.52 introduced a method utilizing the DHq-ROF Aczel-Alsina Weighted Arithmetic (DHq-ROFAAWA) operator and the DHq-ROF Aczel-Alsina Weighted Geometric (DHq-ROFAAWG) operator for the Multi-Criteria Group Decision-Making (MCGDM) process. Additionally, Gayen et al.53 proposed an MCGDM method that employs the DHq-ROF Schweizer-Sklar Weighted Averaging (DHq-ROFSSWA) operator and the DHq-ROF Schweizer-Sklar Weighted Geometric (DHq-ROFSSWG) operator. Since the attribute weights in the above article are given subjectively, the attribute weight vector is assumed to be . Table 15; Fig. 7 illustrate the diagnostic results obtained from diverse diagnostic methods for SD patients. Table 15Comparative analysis of different diagnostic methods in patients with SD.MethodsOperatorsScore valuesRanking The method of Gayen et al.13DHq-ROFAAWA DHq-ROFAAWG The method of Gayen et al.14DHq-ROFSSWA DHq-ROFSSWG Proposed methodDHq-ROFWDHM DHq-ROFWDDHM Comparative analysis of different diagnostic methods in patients with SD. Comparative analysis of different diagnostic methods in patients with SD. Analysis of SD patient diagnostics shows the proposed method’s optimal result () matches existing methods. While other rankings vary slightly, this doesn’t impact the best diagnosis, confirming the method’s reliability. Its advantages will be discussed next. Advantage analysis In this part, we will look into the advantages of the proposed methodology from three perspectives: the comparison of performance in handling expanded fuzzy information spaces, the rate of variation in scoring values among the alternatives, the flexibility and consideration of interrelationships among attributes, the distance formula without normalization, and the overall benefits of the proposed MAGDM approach. Comparison of performance in handling expanded fuzzy information spaces. Comparison of performance in handling expanded fuzzy information spaces. In this section, a comparative analysis is conducted with the methods proposed by Zhou et al.58 and Wang et al.57. Zhou et al.58 developed a MULTIMOORA approach within a Pythagorean hesitant fuzzy environment, which derives the final ranking by integrating the Ratio System Method, the Reference Point Method, and the Full Multiplication Method. Wang et al.57, on the other hand, introduced the hesitant Fermatean fuzzy weighted Bonferroni mean (HFFWBM) operator and the hesitant Fermatean fuzzy Einstein weighted Bonferroni mean (HFFEWBM) operator. For the purpose of this comparison, the attribute weight vector is set as . The comparative results under expanded fuzzy information spaces are shown in Table 16. Table 16Comparative results under expanded fuzzy information spaces.MethodsScore valuesRanking The Pythagorean hesitant fuzzy-MULTIMOORA method of Zhou et al.66 The method of Wang et al.53 HFFWBM HFFEWBM Proposed method DHq-ROFWDHM DHq-ROFWDDHM Comparative results under expanded fuzzy information spaces. As can be seen from Table 16, the best and worst alternatives obtained by Zhou et al.‘s58 method are consistent with those of our proposed method. However, the ranking results for the intermediate alternatives , , and are equal, making it impossible to distinguish their relative order. Furthermore, the Pythagorean hesitant fuzzy sets used by Zhou et al.58 represent a special case of the DHq-ROFSs in our method when the parameter q is set to 2, while the hesitant Fermatean fuzzy sets employed by Wang et al.57 are a special case of the DHq-ROFSs when q is set to 3. Therefore, the method proposed in this paper is capable of representing a broader fuzzy information space. (2)Rate of variation in the scoring values among the alternatives. Rate of variation in the scoring values among the alternatives. It is well known that operators are employed to aggregate fuzzy numbers to ultimately derive score values. A greater difference in these score values indicates a higher capability of the operator, as a larger disparity allows for better differentiation among various alternatives. This paper’s MAGDM approach employs DHq-ROFWDHM and DHq-ROFWDDHM operators. Currently, we compare our results with Gayen et al.52 method. We present Eq. 13 to calculate the overall variation in scoring values (OVSV) and Eq. 14 to calculate the percentage improvement in the overall rate of variation in scoring values (). where represents the variation in the ith scoring value. where represents the of the proposed method and represents the of the compared method. Example 8.1 Since the method attribute weights of Gayen et al.52 are given by experts, the attribute weight vectors are set at the time of comparison as . Table 17 presents the alternative rankings and scores from both Gayen et al.‘s52 method and our proposed approach, while Table 18 displays the obtained . Table 17Comparing the score values obtained using the method developed by Gayen et al. with our own.MethodsOperatorsScore valuesRanking The method of Gayen et al.13DHq-ROFAAWA DHq-ROFAAWG Proposed methodDHq-ROFWDHM DHq-ROFWDDHM Comparing the score values obtained using the method developed by Gayen et al. with our own. Table 18Rate of variation in score values across different Methods.OperatorsVariation in scoring values 1st–2nd2nd–3rd3rd–4th4th–5thDHq-ROFAAWA DHq-ROFWDHM DHq-ROFAAWG DHq-ROFWDDHM Rate of variation in score values across different Methods. Table 17 demonstrates consistent optimal treatment selection despite minor ranking variations between methods. The rate of variation in score values are illustrated in Figs. 8 and 9, clearly indicating that the proposed method demonstrates a greater overall rate of variation in score values. Variation in score values for DHq-ROFAAWA and DHq-ROFWDHM alternatives. Variation in score values for DHq-ROFAAWG and DHq-ROFWDDHM alternatives. From Table 18, it can be concluded that the DHq-ROFWDHM operator shows a 29% improvement compared to the DHq-ROFAAWA operator, while the DHq-ROFWDDHM operator demonstrates a 28% improvement over the DHq-ROFAAWG operator. Both results significantly enhance the rate of variation in score values, indicating that the proposed method is better at distinguishing between alternatives. Hence, compared to the method in Gayen et al.52, the method presented in this paper is more effective in choosing the best alternative. (3)Flexibility and consideration of interrelationships among attributes. Flexibility and consideration of interrelationships among attributes. TCM diagnosis of SD presents a complex challenge due to the variability of patients’ symptoms and the interrelationships among different symptoms. Selecting the appropriate operator is crucial for TCM in diagnosing SD. This paper compares several operators: the DHq-ROFAAWA operator and the DHq-ROFAAWG operator proposed by Gayen et al.52, the HFFWBM operator and the HFFEWBM operator proposed by Wang et al.57. Additionally, the DHq-ROFSSWA operator and the DHq-ROFSSWG operator, also proposed by Gayen et al.53, are examined. Furthermore, Wang et al.56 introduced a MCGDM method utilizing the DHq-ROF Hamacher weighted average (DHq-ROFHWA) operator and the DHq-ROF Hamacher weighted geometric (DHq-ROFHWG) operator. Lastly, Hussain et al.54 proposed a MADM method based on the hesitant q-rung orthopair fuzzy weighted averaging (Hq-ROFWA) operator and the hesitant q-rung orthopair fuzzy weighted geometric (Hq-ROFWG) operator. Meanwhile, Kavitha et al.55 proposed DHq-ROF weighted Dombi arithmetic (DHq-ROFWDA) and DHq-ROF geometric aggregation (DHq-ROFWDG) operators, offering another approach to manage complex and hesitant information. We compare the existing operators with the proposed ones, with results quantitatively presented in Table 19 and visually illustrated in Fig. 10. Graphically, all of them represent the best treatment options; most of the suboptimal alternatives are similar. However, the DHq-ROFAAWA operator proposed by Gayen et al.52 stands out as different. The other alternatives exhibit only slight variations among different operators, but none of these scores significantly impacts the optimal ranking scheme. Table 19Our operators vs. existing operators.MethodsOperatorsScore valuesRanking The method of Gayen et al.13DHq-ROFAAWA DHq-ROFAAWG The method of Wang et al.53HFFWBM HFFEWBM The method of Gayen et al.14DHq-ROFSSWA DHq-ROFSSWG The method of Wang et al.50DHq-ROFHWA DHq-ROFHWG The method of Hussain et al.19Hq-ROFWA Hq-ROFWG The method of Kavitha et al.24DHq-ROFWDA 0.5389 DHq-ROFWDG Proposed methodDHq-ROFWDHM DHq-ROFWDDHM Our operators vs. existing operators. The radar plot of existing operators. From the above, the reliability of the proposed DHq-ROFWDHM operator and the DHq-ROFWDDHM operator can be confirmed. Furthermore, the Dombi norm offers the flexibility to select appropriate parameters based on the patient’s symptoms, while the HM operator considers the interrelationships among attributes. This approach is particularly well-suited for analyzing the correlations between various patient symptoms and the doctor’s diagnosis. As a result, the final ranking may differ slightly from the outcomes produced by the other operators. (4)Distance formula without normalization. Distance formula without normalization. The distance formula serves as a suitable metric for measuring the proximity of two fuzzy numbers. However, existing distance metrics for DHq-ROFNs have certain limitations. For instance, the distance formulas put forward by Kou et al.59 and Wang et al.60 demand an equal quantity of MDs and NMDs. In many practical scenarios, however, the numbers of MDs and NMDs are not equal. Consequently, a complementary strategy must be adopted to ensure that: the number of MDs is identical across compared DHq-ROFNs, and the number of NMDs is likewise identical. Unfortunately, this adjustment can result in a loss of information. Example 8.2 illustrates the superior performance characteristics of our novel distance formula. Example 8.2 Given several DHq-ROFNs: , , , ,, . We compute the distance between the above - mentioned fuzzy numbers with the distance formula from Kou et al.59 and the q-rung orthopair hesitant fuzzy Euclidean distance formula introduced by Wang et al.60. Additionally, we apply our proposed distance formula with the parameter , as well as Wang et al.‘s60 formula with the parameter . The results of these calculations are presented in Table 20. Table 20Our distance formula VS. existing distance formula.Methods and and and NormalizationThe method of Kou et al.26 YesThe method of Wang et al.52 YesProposed method No Our distance formula VS. existing distance formula. As shown in Table 20, the distance between and , calculated using Wang et al.‘s60 distance formula, is 0. However, it is evident that is not equal to . Both Kou et al.59 and Wang et al.60 introduced information loss by employing different normalization methods to ensure that the lengths of DHq-ROFNs are the same. In contrast, our proposed distance formula does not require normalization, thereby avoiding any information loss. (5)Overall advantages of the proposed MAGDM. Overall advantages of the proposed MAGDM. This section systematically evaluates the proposed MAGDM method’s advantages through multi-faceted comparative assessments. In order to emphasize these advantages more distinctly, we summarize several existing state-of-the-art methods (Gayen et al.; Gayen et al.; Zhou et al.; Wang et al.; Hussain et al.; Wang et al.; Kavitha et al.)52–58 in Table 21. Table 21Characteristics of different decision-making methods.MethodsCapturing information by q-ROFSFlexibility due to Dombi operationConsideration of interrelationships between multiple argumentsDistance formula without normalizationObjectively calculating attribute weightsConsidering multiple expertsThe method of Gayen et al.13YesNoNoNoNoYesThe method of Gayen et al.14YesNoNoNoNoYesThe method of Zhou et al.66NoNoNoNoYesNoThe method of Wang et al.53NoNoNoNoNoNoThe method of Wang et al.50YesNoNoNoNoNoThe method of Hussain et al.19YesNoNoNoNoNoThe method of Kavitha et al.24YesYesNoNoNoNoProposed methodYesYesYesYesYesYes Characteristics of different decision-making methods. First and foremost, the proposed method utilizes DHq-ROFSs, which can represent MD and NMD with hesitant values. This approach allows for the adjustment of the q-value, thereby expanding the representation of fuzzy information space. In contrast, Zhou et al.‘s58 methodology employs Pythagorean hesitant fuzzy sets as its foundational framework, which are considered a specific case of DHq-ROFSs when q is set to 2. Similarly, Wang et al.‘s57 methodology employs hesitant Fermatean fuzzy sets as its foundational framework, which are considered a specific case of DHq-ROFSs when q is set to 3. These existing fuzzy sets are unable to effectively address certain cases, while our method can accommodate a broader fuzzy information space through the adjustment of the q-value. Consequently, our method demonstrates greater effectiveness compared to the approaches proposed by Zhou et al.58 and Wang et al.57. Second, this method offers flexibility and the capacity to consider the interrelationships among various attributes. The approach proposed in this paper is founded on the Dombi norm, where the parameter is associated with the DM’s risk preference. The DM can adaptively select the appropriate Dombi parameter based on their individual risk tolerance. Furthermore, in the context of TCM diagnosis of SD, different attributes are interconnected, and the HM operator effectively accounts for these interrelationships. In contrast, the methods presented by (Gayen et al.; Gayen et al.; Hussain et al.; Wang et al.; Wang et al.; Kavitha et al.)52–57 lack these capabilities. Consequently, the proposed methodology yields more reliable and flexible outcomes compared to existing approaches. Third, the method minimizes information loss by introducing a distance metric that does not require normalization. Building on this, the entropy weight method is proposed for objectively calculating attribute weights within the context of DHq-ROFSs, thereby enhancing the reliability of diagnostic results for patients with SD. However, existing methods52–54,56,57 rely on evaluations from experts who assign subjective weights, which undermines the credibility of the diagnostic results. Finally, real-world decision-making inherently involves complexity, and involving multiple experts can provide diverse perspectives and experiences, thereby improving the decision’s scientific validity. (Zhou et al.; Hussain et al.; Wang et al.; Wang et al.; Kavitha et al.)54–58 all focused on the decision-making of a single expert. In contrast, the proposed MAGDM approach accommodates multiple experts, effectively overcoming the limitations of individual decision-making and proving to be more applicable to complex real-world decisions. Conclusions and future research directions In this research, we first propose the DHq-ROFWDHM and DHq-ROFWDDHM operators by integrating HM and Dt-N&t-CNs within the DHq-ROF framework. This integration enhances information aggregation flexibility through adjustable Dombi parameters, and the HM operator effectively captures interrelationships among multiple correlated attributes—this is a key advantage over methods that only handle independent attributes. Second, we introduce a new non-normalized distance measure and an entropy metric for determining unknown attribute weights. The non-normalized distance measure avoids information loss caused by normalization, and the entropy metric eliminates decision-maker subjectivity, which is a critical improvement over subjective weight-assignment methods. Subsequently, we develop a new MAGDM method based on these innovations, and validate its efficacy through a tangible case study on TCM-based sleep disorder diagnosis. Experimental results confirm that our approach, which simultaneously considers flexibility and attribute correlation, achieves a larger variation rate of score values. This strengthens alternative differentiation compared to existing methods, and the approach provides a reliable tool for TCM sleep disorder diagnosis while enriching fuzzy decision-making theory. However, this study has two main limitations. First, while the HM operator captures interrelationships between attributes and addresses the flaw of ignoring attribute dependence, it increases the complexity of attribute fusion. Its aggregation process requires calculating attribute combinations, such as combinations for the -order HM, which consumes more computational resources and reduces efficiency in large-scale tasks. Second, the proposed MAGDM method has not integrated advanced decision-making approaches like TOPSIS, VIKOR, and TODIM. This limits its adaptability to diverse practical scenarios. These limitations may restrict the method’s application efficiency and scope. Future research will focus on addressing current limitations while exploring prospective research directions with targeted innovations. First, adopt advanced fuzzy decision-making methods to streamline HM-based attribute combination calculations, cutting computational costs while preserving attribute correlation capture for large-scale tasks. Second, develop hybrid frameworks by fusing the MAGDM method with advanced approaches—including regret theory and prospect theory, as well as combining it with TODIM to reflect TCM practitioners’ risk preferences or integrating it with VIKOR to balance conflicting attributes like efficacy and safety. Third, extend the DHq-ROF framework to probabilistic linguistic or interval-valued fuzzy environments to handle complex TCM data uncertainty, and verify the optimized method’s reliability and applicability using multi-source to enhance its practical value in clinical decision-making. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Acknowledgements This work was supported by the Gansu Provincial University Teacher Innovation Foundation 2024 A-006, the Key Research and Development Program Project of Gansu Province 24YFFA025, the National Natural Science Foundation of China 12461094 and the National Natural Science Foundation of China 62162055. Author contributions Xiuqin Ma: Writing-Review & Editing, supervision, funding acquisition. Mengyao Wang: Conceptualization, methodology, experimental analysis, validation, writing-original drafts, surveys, data management, visualization. Hongwu Qin: Writing-Review, supervision, funding acquisition. Qi Wei: Check, correctness verification. Tao Li: Check, correctness verification. Funding This work was supported by the Key Research and Development Program Project of Gansu Province 24YFFA025, the National Natural Science Foundation of China 12461094, the Gansu Provincial University Teacher Innovation Foundation 2024 A-006, and the National Natural Science Foundation of China 62162055. Data availability The datasets used and/or analyzed during the current study available from the corresponding author on reasonable request. Declarations Consent for publication Not applicable. Competing interests The authors declare no competing interests. Ethical approval (Not applicable) Consent to participate (Not applicable) Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement The datasets used and/or analyzed during the current study available from the corresponding author on reasonable request.