Individualized dose–response algorithm vs weight-based heparin dosing for intraprocedural anticoagulation during atrial fibrillation ablation Authors: Jared Bryce Pleash, Jonathan O’Leary, Michael David Fryer, Tuppence Richman, Philip Cumpston, Matthew K.Y. Tung Published: 2026 ABSTRACT BackgroundEffective anticoagulation during atrial fibrillation (AF) ablation is essential to minimize thromboembolic and bleeding complications. Conventional weight-based (WB) unfractionated heparin dosing does not account for interindividual variability in anticoagulation response and may result in suboptimal activated clotting time (ACT) control.ObjectiveThis study aimed to compare an individualized dose–response algorithm (DRA) with WB heparin dosing for achieving and maintaining target intraprocedural anticoagulation during AF ablation.MethodsWe performed a retrospective single-center cohort study of 238 consecutive AF ablations (WB n = 125; DRA n = 113). The primary endpoint was anticoagulation precision, defined as variability in deviation from a target ACT of 350 seconds across 4 intraprocedural timepoints. Secondary endpoints included the proportion of ACT measurements of ≥350 seconds and effect modification by weight.ResultsA greater proportion of ACT measurements reached ≥350 seconds at all timepoints in the DRA group than in the WB dosing. Anticoagulation precision was improved with the algorithm, reflected by lower variability in deviation from the target ACT, particularly early in the procedure (standard deviation at 40 minutes 14.3 vs 43.3 seconds; P < .001). Treatment effect on precision was modified by weight (interaction P = .027). Among patients weighing <80 kg, deviation from target was reduced by 26.6 seconds at 10 minutes (P < .001) and 21.1 seconds at 40 minutes (P < .001).ConclusionIndividualized DRA-guided heparin dosing improved anticoagulation precision during AF ablation, with significantly lower variability and more consistent target attainment. Benefits were most pronounced early and among lighter patients. Prospective studies are required to determine whether improved anticoagulation control translates into improved clinical outcomes. Key Findings ▪Individualized dose–response algorithm–guided heparin dosing significantly improved intraprocedural anticoagulation precision compared with conventional weight-based dosing during atrial fibrillation ablation.▪Algorithm-guided dosing resulted in a greater proportion of activated clotting time (ACT) measurements achieving the ≥350-second target at all intraprocedural timepoints.▪The greatest improvements in anticoagulation control were observed early in the procedure, a period associated with the highest thromboembolic risk.▪Treatment effect was modified by body weight, with larger reductions in deviation from target ACT observed in patients weighing <80 kg. Key Findings ▪Individualized dose–response algorithm–guided heparin dosing significantly improved intraprocedural anticoagulation precision compared with conventional weight-based dosing during atrial fibrillation ablation.▪Algorithm-guided dosing resulted in a greater proportion of activated clotting time (ACT) measurements achieving the ≥350-second target at all intraprocedural timepoints.▪The greatest improvements in anticoagulation control were observed early in the procedure, a period associated with the highest thromboembolic risk.▪Treatment effect was modified by body weight, with larger reductions in deviation from target ACT observed in patients weighing <80 kg. ▪Individualized dose–response algorithm–guided heparin dosing significantly improved intraprocedural anticoagulation precision compared with conventional weight-based dosing during atrial fibrillation ablation. Individualized dose–response algorithm–guided heparin dosing significantly improved intraprocedural anticoagulation precision compared with conventional weight-based dosing during atrial fibrillation ablation. ▪Algorithm-guided dosing resulted in a greater proportion of activated clotting time (ACT) measurements achieving the ≥350-second target at all intraprocedural timepoints. Algorithm-guided dosing resulted in a greater proportion of activated clotting time (ACT) measurements achieving the ≥350-second target at all intraprocedural timepoints. ▪The greatest improvements in anticoagulation control were observed early in the procedure, a period associated with the highest thromboembolic risk. The greatest improvements in anticoagulation control were observed early in the procedure, a period associated with the highest thromboembolic risk. ▪Treatment effect was modified by body weight, with larger reductions in deviation from target ACT observed in patients weighing <80 kg. Treatment effect was modified by body weight, with larger reductions in deviation from target ACT observed in patients weighing <80 kg. Introduction Effective anticoagulation is essential for reducing complications associated with atrial fibrillation (AF) ablation procedures, to avoid both thromboembolic and bleeding complications.1, 2, 3 Periprocedural thromboembolic event risks are estimated at 0.15%–0.5%, and pericardial tamponade risks are estimated at 0.4%–1.3%.4 To minimize these risks, guidelines recommend the maintenance of an intraprocedural activated clotting time (ACT) of at least 300 seconds.4 Unfractionated heparin (UFH) dosing has largely relied on weight-based (WB) algorithms, which may inadequately achieve target ACT owing to patient factors apart from weight.5 This has become particularly important after the widespread availability of direct-acting oral anticoagulants (DOACs).6 Multiple studies have demonstrated that different anticoagulants have differing responses to WB doses of heparin.7, 8, 9, 10, 11, 12 Following on from this, tailored approaches to WB heparin dosing, factoring in the type of anticoagulant and baseline ACT, have demonstrated greater efficacy in achieving therapeutic intraprocedural ACT levels.13, 14, 15, 16 An alternative approach for calculating heparin dosage required to reach a target ACT has been developed using personalized estimates of blood volume and in vivo heparin dose–response curves.17,18 In a cohort of 478 patients (using a variety of anticoagulants), this approach was used to achieve a first postheparin ACT goal of 360 seconds.18 A mean ACT of 373 seconds and a median of 355 seconds were achieved with 87.5% of patients achieving a first ACT of >300 seconds. In this cohort, a modeled comparison with a WB approach demonstrated that only 46.2% would have achieved an ACT of >300 seconds using 100 U/kg of heparin.18 These findings suggest that applying individualized dosing algorithms could improve the speed and precision of achieving therapeutic anticoagulation, independent of the patient’s oral anticoagulant regimen. This study sought to compare intraprocedural ACTs achieved using conventional WB heparin dosing with a tailored approach using the heparin dose–response curve algorithm (DRA) based dosing in patients undergoing AF ablation. Methods Study design This was a retrospective single-center cohort study of consecutive patients undergoing AF ablation procedures performed at Sunshine Coast University Hospital, Australia, from July 2019 to November 2024 with a single physician operator. The study was conducted in accordance with the Declaration of Helsinki and was approved by the Gold Coast Hospital and Health Service Human Research Ethics Committee (project no. 91809). Given the retrospective nature of the study and use of deidentified clinical data, the requirement for individual informed consent was waived by the human research ethics committee. A WB dosing strategy was used from July 2019 to May 2022, before changing to a DRA strategy from June 2022 to November 2024. All patients underwent AF ablation on uninterrupted oral anticoagulation, with no changes to institutional protocol throughout the study period. Baseline patient characteristics, intraprocedural ACTs, and requirement for additional intraprocedural heparin boluses were recorded. Procedural target ACT was 350 seconds for the entire cohort. In the WB group, no continuous heparin infusion was used; anticoagulation was managed using intermittent bolus dosing guided by ACT measurements. Patients received an initial WB bolus of UFH (150 IU/kg if on DOAC and 120 IU/kg if on warfarin). ACTs were measured 10 minutes after the first UFH dose and then every 30 minutes during the procedure, with additional boluses administered as required after each measurement. If the ACT was 325–375 seconds, a further bolus of 10% of the initial dose was given. If the ACT was 250–325 seconds, a bolus equivalent to 20% of the initial dose was administered. For ACT values <250 or >375 seconds, additional dosing was at the physician’s discretion. The initial bolus of UFH and the accompanying infusion dose for the DRA group were calculated using the Heparin Heuristics Lite application (Cumpston, Apple Applications Store), which was applied by the procedural physician at the time of the procedure and integrates variables including height, weight, baseline preheparin ACT, sex, and oral anticoagulant type (Figure 1). After the initial calculated bolus dose, a heparin infusion was commenced at the calculated rate. ACTs were measured 10 minutes after infusion commencement and then every 30 minutes thereafter. For ACTs of <350 seconds, additional boluses of UFH were calculated by the application but were administered at the physician’s discretion. The infusions were discontinued once catheters and sheaths were removed from the left atrium or if the ACT was >400 seconds. Baseline (preheparin) ACT was recorded in the DRA group given that this was required for algorithm input; preheparin ACT was not measured in the WB cohort. Although currently implemented as an iOS application, the underlying dosing algorithm is not platform specific and could be adapted for use in other formats, including Web-based or electronic clinical tools.Figure 1Example of dose calculation using the Heparin Heuristics Lite application. A: Entry of patient-specific variables. B: Selection of oral anticoagulant and ACT measurement platform. C: Output demonstrating calculated unfractionated heparin bolus and infusion rate. The panels illustrate the stepwise application of the dose–response algorithm. ACT = activated clotting time. Example of dose calculation using the Heparin Heuristics Lite application. A: Entry of patient-specific variables. B: Selection of oral anticoagulant and ACT measurement platform. C: Output demonstrating calculated unfractionated heparin bolus and infusion rate. The panels illustrate the stepwise application of the dose–response algorithm. ACT = activated clotting time. An example of an algorithm application is presented in Figure 1. Patient-specific variables (height, weight, sex, baseline ACT, and target ACT) were entered, followed by selection of the oral anticoagulant and ACT measurement platform. The algorithm generated an initial UFH bolus and recommended infusion rate, with subsequent ACT measurements used to guide additional bolus dosing. Protamine dosing estimates were also provided if reversal was required. Outcomes The primary outcome was precision, defined as the variability of absolute deviation from the target ACT of 350 seconds. Secondary endpoints included the proportion of ACT measurements ≥350 seconds at each timepoint and assessment of effect modification by weight category. Statistical analysis Generalized estimating equations (GEEs) were used to account for repeated ACT measurements, using an exchangeable correlation structure. The initial model included treatment group, time, age, weight category (≥80 kg), renal function (estimated glomerular filtration rate), sex, and anticoagulation type, with prespecified interactions (treatment-by-time, treatment-by-weight, treatment-by-age). Prespecified covariates were retained if clinically relevant or if their removal worsened model fit. The final GEE model included treatment group, time, weight category (≥80 kg), sex, and prespecified treatment-by-weight and treatment-by-time interactions. Analyses were performed on available data without imputation. Missing ACT measurements were infrequent and not systematically different between groups. Baseline characteristics were compared between groups using Shapiro–Wilk tests, histograms, and quantile–quantile plots to assess normality; approximately normal variables were compared using Student t tests, non-normal variables using Mann–Whitney U tests, and categorical variables using χ2 tests. As a measure of anticoagulation precision, we compared the variability (standard deviation) of ACT deviation between groups at each timepoint and formally tested variance differences using Levene’s test (Brown–Forsythe modification). Model assumptions were evaluated via residual diagnostics and multicollinearity checks, with no violations identified. No adjustment for multiple comparisons was applied to secondary analyses, which should be interpreted as exploratory. All analyses were conducted in Stata 16.1 (StataCorp, College Station, TX). Results Data from 245 consecutive patients were reviewed, with 7 patients excluded owing to incomplete data leaving a total of 238 analyzed (113 in the DRA group and 125 in the WB group). Baseline characteristics were broadly similar between groups (Table 1). Sex distribution differed significantly between groups (male 38.1% DRA vs 59.2% WB; P < .001); however, treatment effects were estimated from models adjusted for sex and other prespecified covariates. The mean preheparin ACT in the DRA group was 162.1 ± 32.4 seconds; a preheparin ACT was not routinely measured in the WB group. The total UFH dose was lower in the DRA group than the WB group (median 20,555 [interquartile range 16,143–24,800] vs 21,998 [interquartile range 17,000–30,397] U; P < .001).Table 1Baseline characteristics of the participantsCharacteristicWB (n = 125)DRA (n = 113)P value∗Age, median [IQR]64 [58–70.5]62 [55–68.5].236Male sex, n (%)74 (59)43 (38)<.001GFR, median [IQR]79 [68.5–90]79 [66–90].619Platelets, median [IQR]228 [197.5–275]241 [197.5–274.5].327Height, median [IQR]175 [167.5–182.75]177 [168–182].726Weight, median [IQR]87 [72.5–99]88 [77.5–102].278BMI, median [IQR]28.3 [25.2–31.4]29.0 [25.7–32.3].29Rivaroxaban, n (%)73 (58.4)64 (56.6).934Apixaban, n (%)46 (36.8)43 (38.0)Dabigatran, n (%)4 (3.2)3 (2.7)Warfarin, n (%)2 (1.6)3 (2.7)BMI = body mass index; DRA = dose–response curve algorithm group; IQR = interquartile range; GFR = glomerular filtration rate; WB = weight-based group.∗Values are considered significant if P < .05. Table 1Baseline characteristics of the participantsCharacteristicWB (n = 125)DRA (n = 113)P value∗Age, median [IQR]64 [58–70.5]62 [55–68.5].236Male sex, n (%)74 (59)43 (38)<.001GFR, median [IQR]79 [68.5–90]79 [66–90].619Platelets, median [IQR]228 [197.5–275]241 [197.5–274.5].327Height, median [IQR]175 [167.5–182.75]177 [168–182].726Weight, median [IQR]87 [72.5–99]88 [77.5–102].278BMI, median [IQR]28.3 [25.2–31.4]29.0 [25.7–32.3].29Rivaroxaban, n (%)73 (58.4)64 (56.6).934Apixaban, n (%)46 (36.8)43 (38.0)Dabigatran, n (%)4 (3.2)3 (2.7)Warfarin, n (%)2 (1.6)3 (2.7)BMI = body mass index; DRA = dose–response curve algorithm group; IQR = interquartile range; GFR = glomerular filtration rate; WB = weight-based group.∗Values are considered significant if P < .05. Baseline characteristics of the participants BMI = body mass index; DRA = dose–response curve algorithm group; IQR = interquartile range; GFR = glomerular filtration rate; WB = weight-based group. Values are considered significant if P < .05. The dispersion of raw ACT values at each timepoint, reflecting overall anticoagulation variability, is presented in Table 2. Importantly, the DRA group showed markedly lower variability in absolute ACT values, with standard deviations approximately half those of the WB group at early timepoints.Table 2Mean ACT and standard deviation by treatment group and timepointTimepointGroupMean ACT (s)Standard deviation (s)ACT1 (10 min)WB336.1853.84DRA360.0830.54ACT2 (40 min)WB326.2653.16DRA362.7521.13ACT3 (70 min)WB333.7948.26DRA366.3222.84ACT4 (100 min)WB340.1734.95DRA370.8617.33ACT = activated clotting time; ACT1 = 10 minutes after the initial dose of unfractionated heparin; ACT2 = 40 minutes after the initial dose of unfractionated heparin; ACT3 = 70 minutes after the initial dose of unfractionated heparin; ACT4 = 100 minutes after the initial dose of unfractionated heparin; DRA = dose–response curve algorithm; WB = weight-based. Table 2Mean ACT and standard deviation by treatment group and timepointTimepointGroupMean ACT (s)Standard deviation (s)ACT1 (10 min)WB336.1853.84DRA360.0830.54ACT2 (40 min)WB326.2653.16DRA362.7521.13ACT3 (70 min)WB333.7948.26DRA366.3222.84ACT4 (100 min)WB340.1734.95DRA370.8617.33ACT = activated clotting time; ACT1 = 10 minutes after the initial dose of unfractionated heparin; ACT2 = 40 minutes after the initial dose of unfractionated heparin; ACT3 = 70 minutes after the initial dose of unfractionated heparin; ACT4 = 100 minutes after the initial dose of unfractionated heparin; DRA = dose–response curve algorithm; WB = weight-based. Mean ACT and standard deviation by treatment group and timepoint ACT = activated clotting time; ACT1 = 10 minutes after the initial dose of unfractionated heparin; ACT2 = 40 minutes after the initial dose of unfractionated heparin; ACT3 = 70 minutes after the initial dose of unfractionated heparin; ACT4 = 100 minutes after the initial dose of unfractionated heparin; DRA = dose–response curve algorithm; WB = weight-based. Precision, defined as variability of absolute deviation from the target ACT of 350 seconds, is presented in Table 3. In contrast to raw ACT dispersion (Table 2), deviation-based variability more directly captures consistency of target attainment. Variability was consistently lower with the algorithm, particularly at earlier timepoints as seen in Table 3 and Figure 2. Standard deviations were lower with the algorithm at each timepoint. Levene’s tests (Brown–Forsythe) confirmed significantly reduced variance for the algorithm at timepoints 1–3 and a nonsignificant trend at timepoint 4.Table 3Variability in ACT deviation by treatment group and timepointTimepointGroupSD (s)Levene’s P value∗ACT1 (10 min)WB37.6<.001DRA18.8ACT2 (40 min)WB43.3<.001DRA14.3ACT3 (70 min)WB41.0.021DRA16.7ACT4 (100 min)WB25.9.059DRA15.7Variability expressed as the SD of the absolute deviation from the target ACT of 350 seconds.ACT = activated clotting time; ACT1 = 10 minutes after the initial dose of unfractionated heparin; ACT2 = 40 minutes after the initial dose of unfractionated heparin; ACT3 = 70 minutes after the initial dose of unfractionated heparin; ACT4 = 100 minutes after the initial dose of unfractionated heparin; DRA = dose–response curve algorithm; SD = standard deviation; WB = weight-based.∗Values are considered significant if P < .05.Figure 2Variability in ACT deviation by treatment group and timepoint. Variability expressed as the standard deviation of absolute deviation from the target ACT of 350 seconds. ACT = activated clotting time; ACT1 = 10 minutes after the initial dose of unfractionated heparin; ACT2 = 40 minutes after the initial dose of unfractionated heparin; ACT3 = 70 minutes after the initial dose of unfractionated heparin; ACT4 = 100 minutes after the initial dose of unfractionated heparin; DRA = dose–response curve algorithm; WB = weight-based. Table 3Variability in ACT deviation by treatment group and timepointTimepointGroupSD (s)Levene’s P value∗ACT1 (10 min)WB37.6<.001DRA18.8ACT2 (40 min)WB43.3<.001DRA14.3ACT3 (70 min)WB41.0.021DRA16.7ACT4 (100 min)WB25.9.059DRA15.7Variability expressed as the SD of the absolute deviation from the target ACT of 350 seconds.ACT = activated clotting time; ACT1 = 10 minutes after the initial dose of unfractionated heparin; ACT2 = 40 minutes after the initial dose of unfractionated heparin; ACT3 = 70 minutes after the initial dose of unfractionated heparin; ACT4 = 100 minutes after the initial dose of unfractionated heparin; DRA = dose–response curve algorithm; SD = standard deviation; WB = weight-based.∗Values are considered significant if P < .05. Variability in ACT deviation by treatment group and timepoint Variability expressed as the SD of the absolute deviation from the target ACT of 350 seconds. ACT = activated clotting time; ACT1 = 10 minutes after the initial dose of unfractionated heparin; ACT2 = 40 minutes after the initial dose of unfractionated heparin; ACT3 = 70 minutes after the initial dose of unfractionated heparin; ACT4 = 100 minutes after the initial dose of unfractionated heparin; DRA = dose–response curve algorithm; SD = standard deviation; WB = weight-based. Values are considered significant if P < .05. Variability in ACT deviation by treatment group and timepoint. Variability expressed as the standard deviation of absolute deviation from the target ACT of 350 seconds. ACT = activated clotting time; ACT1 = 10 minutes after the initial dose of unfractionated heparin; ACT2 = 40 minutes after the initial dose of unfractionated heparin; ACT3 = 70 minutes after the initial dose of unfractionated heparin; ACT4 = 100 minutes after the initial dose of unfractionated heparin; DRA = dose–response curve algorithm; WB = weight-based. The proportion of intraprocedural ACT measurements achieving ≥350 seconds at each timepoint is presented in Table 4. At all 4 measurements, a significantly greater proportion of ACTs in the DRA group met or exceeded the 350-second threshold compared with WB dosing. Mean ACT values subsequently remained below the 350-second threshold in the WB group but exceeded this throughout each timepoint in the DRA group, as seen in Figure 3.Table 4Proportion of an ACT of ≥350 seconds by group and timepointTimepointWB ≥350, n (%)DRA ≥350, n (%)P value∗ACT1 (10 min)72 (57.6)86 (76.1).003ACT2 (40 min)54 (43.6)85 (77.3)<.001ACT3 (70 min)47 (44.3)72 (82.8)<.001ACT4 (100 min)46 (59.0)33 (94.3)<.001ACT = activated clotting time; ACT1 = 10 minutes after the initial dose of unfractionated heparin; ACT2 = 40 minutes after the initial dose of unfractionated heparin; ACT3 = 70 minutes after the initial dose of unfractionated heparin; ACT4 = 100 minutes after the initial dose of unfractionated heparin; DRA = dose–response curve algorithm; n = number of ACT measurements meeting or exceeding 350-second threshold; WB = weight-based.∗Values are considered significant if P < .05.Figure 3Mean ACT over time. ACT = activated clotting time; ACT1 = 10 minutes after the initial dose of unfractionated heparin; ACT2 = 40 minutes after the initial dose of unfractionated heparin; ACT3 = 70 minutes after the initial dose of unfractionated heparin; ACT4 = 100 minutes after the initial dose of unfractionated heparin; DRA = dose–response curve algorithm; WB = weight-based. Table 4Proportion of an ACT of ≥350 seconds by group and timepointTimepointWB ≥350, n (%)DRA ≥350, n (%)P value∗ACT1 (10 min)72 (57.6)86 (76.1).003ACT2 (40 min)54 (43.6)85 (77.3)<.001ACT3 (70 min)47 (44.3)72 (82.8)<.001ACT4 (100 min)46 (59.0)33 (94.3)<.001ACT = activated clotting time; ACT1 = 10 minutes after the initial dose of unfractionated heparin; ACT2 = 40 minutes after the initial dose of unfractionated heparin; ACT3 = 70 minutes after the initial dose of unfractionated heparin; ACT4 = 100 minutes after the initial dose of unfractionated heparin; DRA = dose–response curve algorithm; n = number of ACT measurements meeting or exceeding 350-second threshold; WB = weight-based.∗Values are considered significant if P < .05. Proportion of an ACT of ≥350 seconds by group and timepoint ACT = activated clotting time; ACT1 = 10 minutes after the initial dose of unfractionated heparin; ACT2 = 40 minutes after the initial dose of unfractionated heparin; ACT3 = 70 minutes after the initial dose of unfractionated heparin; ACT4 = 100 minutes after the initial dose of unfractionated heparin; DRA = dose–response curve algorithm; n = number of ACT measurements meeting or exceeding 350-second threshold; WB = weight-based. Values are considered significant if P < .05. Mean ACT over time. ACT = activated clotting time; ACT1 = 10 minutes after the initial dose of unfractionated heparin; ACT2 = 40 minutes after the initial dose of unfractionated heparin; ACT3 = 70 minutes after the initial dose of unfractionated heparin; ACT4 = 100 minutes after the initial dose of unfractionated heparin; DRA = dose–response curve algorithm; WB = weight-based. In GEE models accounting for repeated measures, the algorithm was associated with significantly lower deviation from the 350-second target. There was evidence of interaction between treatment and weight category (Wald χ2 test for interaction term, P = .027), with larger reductions in deviation observed among patients weighing <80 kg, particularly at earlier timepoints. The estimated effect of the DRA strategy compared with WB dosing on deviation from the ACT target at each timepoint stratified by weight group is presented in Table 5. Values represent marginal effects from the adjusted GEE model, with negative values indicating reduced deviation from target (improved precision).Table 5Effect of algorithm vs weight-based dosing on deviation from target ACTWeight groupTimepointEffect of DRA vs WB (s)95% confidence intervalP value∗<80 kgACT1 (10 min)−26.59(−40.16 to −13.02)<.001ACT2 (40 min)−21.10(−30.72 to −11.48)<.001ACT3 (70 min)−14.63(−34.55 to 5.29).150ACT4 (100 min)−9.33(−27.30 to 8.64).309≥80 kgACT1 (10 min)−8.42(−17.02 to 0.19).055ACT2 (40 min)−18.38(−29.73 to −7.03).001ACT3 (70 min)−3.51(−10.77 to 3.74).342ACT4 (100 min)1.71(−5.79 to 9.20).656ACT = activated clotting time; ACT1 = 10 minutes after the initial dose of unfractionated heparin; ACT2 = 40 minutes after the initial dose of unfractionated heparin; ACT3 = 70 minutes after the initial dose of unfractionated heparin; ACT4 = 100 minutes after the initial dose of unfractionated heparin; DRA = dose–response curve algorithm; WB = weight-based.∗Values are considered significant if P < .05. Table 5Effect of algorithm vs weight-based dosing on deviation from target ACTWeight groupTimepointEffect of DRA vs WB (s)95% confidence intervalP value∗<80 kgACT1 (10 min)−26.59(−40.16 to −13.02)<.001ACT2 (40 min)−21.10(−30.72 to −11.48)<.001ACT3 (70 min)−14.63(−34.55 to 5.29).150ACT4 (100 min)−9.33(−27.30 to 8.64).309≥80 kgACT1 (10 min)−8.42(−17.02 to 0.19).055ACT2 (40 min)−18.38(−29.73 to −7.03).001ACT3 (70 min)−3.51(−10.77 to 3.74).342ACT4 (100 min)1.71(−5.79 to 9.20).656ACT = activated clotting time; ACT1 = 10 minutes after the initial dose of unfractionated heparin; ACT2 = 40 minutes after the initial dose of unfractionated heparin; ACT3 = 70 minutes after the initial dose of unfractionated heparin; ACT4 = 100 minutes after the initial dose of unfractionated heparin; DRA = dose–response curve algorithm; WB = weight-based.∗Values are considered significant if P < .05. Effect of algorithm vs weight-based dosing on deviation from target ACT ACT = activated clotting time; ACT1 = 10 minutes after the initial dose of unfractionated heparin; ACT2 = 40 minutes after the initial dose of unfractionated heparin; ACT3 = 70 minutes after the initial dose of unfractionated heparin; ACT4 = 100 minutes after the initial dose of unfractionated heparin; DRA = dose–response curve algorithm; WB = weight-based. Values are considered significant if P < .05. Discussion This study provides the first comparison between conventional WB heparin dosing and a tailored DRA approach to heparin dosing for patients undergoing AF ablation. In our single-center cohort of consecutive patients, we found that the DRA approach achieved ACT values closer to a 350-second target with less variability (greater precision), particularly early in the procedure. On top of this, the DRA approach led to a higher proportion of patients achieving an ACT of ≥350 seconds. Notably, improved ACT precision and a higher proportion of ACT measurements of ≥350 seconds were achieved without increased heparin dosing and with a modest reduction in total heparin administered. This suggests that the algorithm improved dosing efficiency rather than simply increasing anticoagulant exposure. Although mean ACTs for both strategies generally approximated the 350-second target at each timepoint, the algorithm reduced variability both in absolute ACT values (Table 2) and in deviation from the target ACT (Table 3), indicating more consistent and predictable anticoagulation. The algorithm’s benefit was most pronounced in the critical early phase of the procedure when thromboembolic risk is highest. Lighter patients (<80 kg) experienced larger reductions in absolute deviation at the first 2 measurements, whereas heavier patients (≥80 kg) derived a smaller early benefit and a significant improvement mainly at timepoint 2. These patterns suggest that the algorithm improves initial target attainment and early maintenance, with potential ceiling effects or reduced marginal returns among heavier patients as the case progresses. Mechanistically, several factors may explain the observed effects. The algorithm incorporates baseline ACT, body size (height, weight), sex, estimated blood volume, and concomitant anticoagulant type, variables known to influence heparin pharmacodynamics and the ACT response.5,7, 8, 9,12,17, 18, 19, 20, 21 By tailoring the initial bolus and infusion rate, the algorithm likely reduces under- and overdosing that can occur with fixed WB protocols, particularly in patients on DOACs where ACT responses to UFH vary by agent and timing of the last dose.8, 9, 10, 11, 12 The attenuation of benefit at later timepoints may reflect stabilization of ACT with repeated titration under both strategies, operator discretion overriding the algorithm when ACTs approach high values, or diminishing returns once most patients are near target. In heavier patients, higher circulating blood volume and distribution kinetics could blunt incremental gains from standardized adjustment steps, pointing to a need for weight-adjusted refinement of later dosing logic. Clinically, improved precision (narrower ACT dispersion) may be more meaningful than small differences in mean ACT. Time within the desired ACT range is associated with a lower risk of thromboembolic events and potentially fewer bleeding complications from extreme ACT outliers.1, 2, 3, 4 Although our study did not adjudicate clinical endpoints (eg, stroke, transient ischemic attack, tamponade, major bleeding), the substantial reduction in variability with the algorithm suggests a more reliable anticoagulation profile that could translate into safer procedures, fewer emergent corrections, and more efficient workflow. Our findings align with previous work showing that individualized dosing strategies, incorporating baseline ACT, anticoagulant class, and patient-specific parameters, improve the likelihood of achieving therapeutic ACT early in the procedure and reduce dosing volatility.13, 14, 15, 16,18 Earlier cohorts have demonstrated that fixed WB boluses often underachieve ACT targets in patients on DOACs, whereas individualized or algorithmic approaches improve first-measure target attainment.7, 8, 9, 10,12, 13, 14, 15, 16 The present analysis extends those observations by quantifying precision over serial timepoints and demonstrating effect modification by weight, thereby informing where algorithmic dosing adds the most value and where it may require further refinement. Study limitations This study has important limitations. Its retrospective, single-center, single-operator design introduces potential selection, performance, and temporal confounding, especially because the 2 strategies were used in 2 different time periods (before/after algorithm implementation). We cannot exclude secular changes in case mix, ablation approaches, and catheter technology that might influence ACT profiles independently of dosing strategy. Despite these limitations, the WB and DRA cohorts appeared broadly comparable across baseline characteristics that would likely affect intraprocedural ACTs. Baseline ACT was incorporated into the DRA but was not measured in the WB group, and therefore, we cannot exclude the possibility that unmeasured baseline differences contributed in part to observed early ACT differences. In addition, the DRA incorporates the Nadler formula to estimate blood volume, which may exhibit proportional bias at higher body weights, potentially reducing dosing accuracy in heavier patients.22,23 ACT measurement devices themselves may also demonstrate proportional bias, particularly at higher ACT values, which could further influence anticoagulation assessment.24, 25, 26 These factors may partially explain the attenuated benefit of the algorithm observed in patients weighing ≥80 kg and highlight areas for refinement in individualized dosing and monitoring approaches. The study was also not powered or designed to assess clinical endpoints such as procedural complications. Given the low incidence of clinically important events such as stroke and tamponade, ACT represents a pragmatic and widely accepted surrogate endpoint used in previous studies evaluating anticoagulation strategies.13, 14, 15, 16 Finally, we did not capture procedural duration, number of corrective heparin boluses, or cost-related data, limiting our ability to assess procedural efficiency, resource utilization, and broader patient-centered or operational benefits. Future prospective, randomized multicenter trials with larger cohorts would be able to address these study limitations and further validate these findings. These trials could assess time-in-target range, number and size of corrective boluses, procedure efficiency, and clinical endpoints (stroke/transient ischemic attack, pericardial tamponade, major bleeding). Prespecified subgroup analyses by anticoagulant class, renal function, body mass index, and baseline ACT could clarify for whom the algorithm offers the greatest benefit. Cost-effectiveness and usability studies would also be important for broader adoption. Conclusion Individualized DRA-guided heparin dosing improved the precision of intraprocedural anticoagulation during AF ablation, with the largest gains early in the procedure and greater benefits among lighter patients. Although mean ACTs between strategies were similar overall, reduced variability indicates more consistent target attainment. Given the study’s nonrandomized design and lack of clinical outcomes, these results should be considered hypothesis generating. Prospective, comparative studies are warranted to determine whether improved ACT precision translates into fewer thromboembolic and bleeding events, more efficient workflow, and better overall procedural outcomes. Disclosures Dr Philip Cumpston has published the iPhone application Heparin Heuristics Lite, which was used to assist in estimating heparin doses during this study. No other authors report conflicts of interest relevant to this work.