Impact of wear position on dosimeter performance: measurement validity under simulated indoor illumination Authors: S W de Vries, J Mardaljevic, J van Duijnhoven Published: 2026 ABSTRACT AbstractThe light an individual is exposed to, or personal light exposure (PLE), plays an important role in human health and well-being research. PLE is typically quantified using body-worn light dosimeters (sensors), but their effectiveness as a proxy for measuring at the eye remains insufficiently understood. By combining 3D body scans and lighting simulations, we quantified the discrepancies between chest-worn and eye-level dosimeter measurements under three generalized indoor lighting scenarios. Chest-worn dosimeter performance was found to be highly context-dependent, with the smallest deviations (between −26.2% and 63.8%) observed for devices on the lower chest. Quantifying corneal light exposure, i.e., considering the field of view of the human eye, based on chest-worn dosimeter measurements proved problematic as deviations under overhead lighting exceeded a minimum 25%. To deepen our understanding of the underlying mechanisms, we disaggregated the causes of these discrepancies into three factors: the translational dosimeter displacement, the rotational dosimeter displacement, and body self-occlusion. The results indicate that minimizing the rotational displacement is key to reducing measurement discrepancies and inter-individual variability. To improve the validity of chest-worn dosimetry, an individualized approach is recommended: selecting a chest position where the dosimeter best aligns with the subject’s typical view direction during expected activities.Subject terms: Engineering, Health care, Medical research, Optics and photonics, Physics Introduction Light has a profound impact on various physiological and behavioral processes related to human health and well-being, known as the non-image-forming (NIF) effects of light1–5. To investigate such effects in real-world environments, recordings of the light an individual is exposed to over time, often referred to as personal light exposure (PLE), can provide valuable insights. Given that humans can experience vastly different illumination fields6 and that their light exposure is affected by their behavior7, PLE must be measured on an individual basis8–10. PLE is typically recorded using small wearable light sensors known as lightloggers, light dosimeters, or, more concisely, dosimeters. Dosimeters are used in studies with various objectives, for example, to quantify inter-individual differences in PLE6,11,12, to investigate the role of PLE in the development of myopia13,14, and to explore associations between PLE and behavioral and physiological outcomes related to NIF effects, for example, sleep15–21, alertness15,22–24, and mood16,17. Collectively, such studies are generally referred to as light-dosimetry field studies. In these contexts, the physiologically relevant light exposure is at the eye. However, mainly for reasons of practicality, dosimeters are often worn elsewhere on the body, typically on the chest or wrist25. The validity of using such proxy positions to measure eye-level light exposure has been an ongoing question for the last few decades, with the earliest study being reported in 198326. Since then, multiple studies have sought to address this issue by comparing dosimeter measurements taken at these proxy positions on the body and the eye27–38. However, the reported correlations and discrepancies between body-worn and eye-level measurements vary across studies, with conflicting claims regarding the performance of body-worn dosimeters for predicting eye-level light exposure. Generally, these studies suggest that measurements from chest-worn dosimeters approximate eye-level light exposure more closely than those from wrist-worn devices29,35–37, although this finding is not observed across all studies34,38. A primary reason for the competing claims in these studies is that they were mostly conducted under naturalistic (real-world) conditions, rather than in controlled environments. As a result, the types of illumination fields under which the measurements were performed, as well as the typical activities (i.e., postures) of their subjects, varied across studies. Depending on these contextual factors, the amount of light measured at a certain position on the body can differ considerably31,35. For example, under a relatively uniform illumination field, unobstructed dosimeters at different positions on the body tend to measure similar light levels, even when their orientations vary substantially. In contrast, under directional illumination, such as that produced by overhead luminaires in offices, small differences in the orientation of the dosimeter can lead to large variations in the measured illumination. Additionally, the use of a dosimeter necessarily introduces uncertainty in the results due to limitations in the device’s intrinsic photometric sensor performance36,39. Taken together, the use of real-world conditions such as those used in the aforementioned studies makes it challenging to draw firm conclusions on the in-situ performance of body-worn dosimeters, i.e., the extent to which the measured illumination accurately represents the intended measure, in this case, light exposure at the eye. To overcome the limitations of real-world methodologies, we devised a virtual environment to investigate the in-situ performance of body-worn dosimeters using a hybrid measurement-simulation approach that combines high-resolution 3D body scans and lighting simulation techniques. In the preceding study40, we identified three in-situ factors that are likely to be the major causes of the discrepancies between body-worn and eye-level measurements. The three factors are: the translational dosimeter displacement (TDD); rotational dosimeter displacement (RDD); and body self-occlusion (BSO). These factors are illustrated in Fig. 1 using one of the subjects used in the study. Based on body scans of twelve subjects in three postures, individual TDD, RDD, and BSO factors were quantified using the Radiance lighting simulation system, widely regarded as the most rigorously validated tool to predict physically-based luminous quantities under daylight and artificial lighting41–43. It was revealed that minor shifts in dosimeter placement can lead to substantial variations in the magnitude of these three factors40. TDD and RDD are purely geometrical factors, whereas BSO was determined for a perfectly diffuse luminous environment – in effect, an ideal integrating sphere. Exactly how these factors combine to cause the discrepancy between a body-worn dosimeter measurement and that recorded at the eye depends on both the properties of the illumination environment and the unique body morphology of the individual. Three in-situ factors impacting dosimeter performance. The a translational dosimeter displacement (TDD), b rotational dosimeter displacement (RDD), and c body self-occlusion (BSO) for a dosimeter at position P with sensor normal np, relative to an `ideal' dosimeter at the eye point (Peye) with view direction (normal) D. α is the angle between D and np. Figure and caption reproduced from the preceding study40. In the study described here, we demonstrate how the TDD, RDD, and BSO factors combine to impact the performance of body-worn dosimetry. We do this by quantifying the differences between body-worn and eye-level dosimeter measurements under three non-uniform illumination conditions. Because we expect the nature of each particular illumination condition to be a significant factor in the evaluation, we devised three contextual lighting scenarios based on generalized representations of typical indoor lighting. Lastly, to note, in this article, light exposure is used as a generic term for a series of light measurements and thus not as a synonym for the photometric quantity luminous exposure (Hv). These measurements can be in any quantity, e.g., photopic illuminance or melanopic equivalent daylight illuminance (EDI)44. Similarly, to refer to a single measurement, in any quantity, the term light level is used. Results The in-situ performance of chest-worn dosimeters for predicting eye-level light exposure is presented based on three generalized indoor lighting scenarios. First, an illustrative example is provided to explain the structure and interpretation of the results. This is followed by three figures: the first shows dosimeter performance across subjects; the second examines the contributions of the TDD, RDD, and BSO factors to the observed differences; and the third illustrates the extent of inter-individual variability in dosimeter performance among our subjects. Illustration of the results To introduce the structure of the results, we first present a subset of the simulation data to illustrate the format of the plots. Figure 2a shows the simulation results for the twelve seated subjects who were writing (desk-facing). On the y-axis, the relative differences D between the simulated measurements of the dosimeter on the upper-central chest relative to the eye-level dosimeter (with, in this case, a hemispherical field of view (FOV)) are given. A positive difference indicates that the chest-worn dosimeter overpredicts actual light levels at the eyes. On the x-axis, the seven room positions are given. The three considered lighting scenarios, i.e., diffuse top lighting, directional top lighting, and diffuse side lighting, are indicated in black, gray, and red, respectively. The presented deviations are thus strictly scenario-dependent bounds. For each subject, the results across the room positions are connected by a line, resulting in twelve lines per lighting scenario. Example plots to illustrate the presentation of the results. a The relative differences (Dc) between simultaneously measured light levels (irradiances) at the upper-central chest and eyes of all twelve subjects. A positive value indicates that the chest-worn dosimeter overpredicts actual light levels at the eyes. b The corresponding across-subject means (D¯c), i.e., the mean differences for each room position, averaged over all participants. c The corresponding within-subject means (D~c), i.e., the mean differences for each subject, averaged over all room positions. Based on the data presented in this plot, the mean relative differences across (D¯c) and within subjects (D~c) were calculated. The mean differences across all subjects (D¯c), for each position in the room, separately for each lighting scenario, are shown in Fig. 2b. Each scenario is indicated by a line, and a cross indicates the room position nearest the front of the room (the 1 m position), while the subsequent six positions are indicated by full circles. To present the data efficiently, the results were stacked vertically. The mean differences within subjects (D~c) are shown using circles in Fig. 2c, resulting in twelve circles per lighting scenario. For each scenario, a boxplot was constructed based on these individual subject means to illustrate inter-individual variability in the measurements. Chest-worn dosimeter performance across subjects The mean relative differences across subjects (D¯c) between light levels measured by the four chest-worn dosimeters and the eye-level dosimeter are given in Fig. 3. A broken y-axis is used to accommodate all values, while preserving detail in the lower range. In each panel, the left side indicates the difference relative to an eye-level dosimeter with a standard hemispherical FOV, and the right side indicates the difference relative to an eye-level dosimeter with human FOV occlusion. For all postures, relative to a dosimeter at the eyes with a hemispherical FOV, the mean relative difference is between −19.8% and 82.6% for a dosimeter worn on the upper chest and between −26.2% and 63.8% for a dosimeter worn on the lower chest. When incorporating FOV occlusion for the dosimeter at the eyes, the range of possible differences is considerably larger, at −18.6% to 658.2% for the upper chest and −24.0% to 564.4% for the lower chest. This increased range is mainly attributable to the results of the subjects who were seated and looking toward the desk under the diffuse side lighting scenario. When not considering this specific case, the possible differences for a dosimeter on the upper chest range from −18.6% to 141.2%, and for a dosimeter on the lower chest from −24.0% to 117.0%. Performance of chest-worn dosimeters across subjects. Mean relative differences (D¯c) between light levels (irradiances) measured by four chest-worn dosimeters and a dosimeter at the eyes with a standard hemispherical FOV and with human FOV occlusion (defined by CIE S 026:201844). A positive value indicates that the chest-worn dosimeter overpredicts actual light levels at the eyes. In Fig. 4, the mean relative differences (D¯c) under the three lighting scenarios are again shown using a black, gray, or red circle. These are the same values as presented on the left side of each panel in Fig. 3. Here, we show how the individual contributions of the TDD, RDD, and BSO factors contribute to each ‘total’ mean relative difference. For each factor, again a cross indicates the contribution at the position nearest the front of the room, whereas a square indicates the other six positions. The sum of these three contributions equals the mean relative difference. The boxes indicate to which mean relative difference the contributions correspond. Generally, RDD results in a positive difference, while TDD and BSO lead to negative differences. Impact of three in-situ factors on chest-worn dosimeter performance. Contributions of translational dosimeter displacement (TDD), rotational dosimeter displacement (RDD), and body self-occlusion (BSO) factors to the mean relative differences (D¯c) between light levels (irradiances) measured by chest-worn dosimeters and a dosimeter at the eyes with a standard hemispherical FOV. A positive value indicates that the chest-worn dosimeter overpredicts actual light levels at the eyes. For each scenario, the sum of the three in-situ factors (displayed within a box) is equal to the `total' mean relative difference. Inter-individual variability in chest-worn dosimeter performance The inter-individual variability in chest-worn dosimeter measurements is illustrated in Fig. 5 using boxplots of the mean relative differences of each subject (D~c). As in Fig. 3, a broken y-axis is used, and the left side of each panel indicates the differences relative to an eye-level dosimeter with a hemispherical FOV, the right side to a dosimeter with human FOV occlusion. The range of inter-individual variability increases considerably depending on the subjects’ posture. Relative to an eye-level dosimeter with a hemispherical FOV, the maximum observed interquartile range spans 36.1% for the standing and seated, screen-facing posture, and 76.9% for the seated, desk-facing posture. When compared to an eye-level dosimeter with FOV occlusion, the maximum interquartile range increases to 56.3% for standing and seated, screen-facing postures, and to 224.2% for the seated, desk-facing posture. Inter-individual variability in chest-worn dosimeter performance. Inter-individual variability in the mean relative differences (D~c) between light levels (irradiances) measured by four chest-worn dosimeters and a dosimeter at the eyes with a standard hemispherical FOV and with human FOV occlusion (defined by CIE S 026:201844). A positive value indicates that the chest-worn dosimeter overpredicts actual light levels at the eyes. Discussion By combining 3D body scans of twelve subjects with lighting simulations, we quantified the range of possible deviations between chest-worn and eye-level dosimeter measurements under three generalized indoor lighting scenarios. To identify the underlying causes of these deviations, we disaggregated the results to determine the effect of three in-situ factors impacting dosimeter performance: TDD, RDD, and BSO. It is essential to acknowledge that the three lighting scenarios we employed are idealized yet representative of typical indoor illumination fields. The presented deviations (Fig. 3) under each illumination field are thus scenario-dependent bounds, and hence should be interpreted with caution. Taken together, the range of measurement deviations under these scenarios offers a reasonable indication of the range one might expect when using a chest-worn dosimeter to approximate eye-level measurements in indoor environments. The exact measurement deviation, i.e., the value within this range, will depend on the composition of the illumination field. While we observed measurement deviations between dosimeters positioned on the upper and lower chest, there were generally no substantial differences between dosimeters placed on the lateral and central chest. The measurement deviations computed at each of the seven room positions were fairly consistent, except for the point nearest the front of the room (1 m position, indicated by a cross in Fig. 3). This inconsistency is mainly due to less of the ceiling emitter being visible at this position, as well as being very close to the edge of the vertical emitter. Across the three lighting scenarios, the range of mean relative differences between chest-worn and eye-level measurements (left sides of the panels in Fig. 3) was approximately −19.8% to 82.6% for dosimeters on the upper chest, and −26.2% to 63.8% for dosimeters on the lower chest. In general, for subjects in the standing and seated, screen-facing postures, chest-worn dosimeters overpredict eye-level light exposure under top lighting and underpredict under side lighting. Since the range of possible deviations under top lighting was greater (overprediction up to 82.6%) than under side lighting (underprediction up to −26.2%), achieving mean relative differences close to zero would require the lighting in a space to have a greater contribution of side lighting relative to top lighting. This does not apply to subjects in a seated, desk-facing posture, as then the chest-worn dosimeter overpredicts eye-level light exposure under all lighting scenarios, underscoring that the deviations are strongly dependent on subject posture. Considering chest-worn dosimeter measurements relative to an eye-level dosimeter with human FOV occlusion (right sides of the panels in Fig. 3), the most striking deviations appear for the seated, desk-facing posture under diffuse side lighting. In this condition, deviations of up to 658.2% are present, indicating considerable overprediction. When a subject looks toward the desk, an eye-level dosimeter with a hemispherical FOV retains a direct view of the vertical emitter at the upper edge of its FOV. With FOV occlusion, however, this direct line of sight is occluded, and only light reflected by the environment reaches the eye-level dosimeter. By contrast, the chest-worn dosimeter, due to its more upward orientation (Fig. 6), maintains a direct line of sight to the emitter. Since light reaching the back of the room undergoes multiple reflections and therefore diminishes in radiant intensity relative to direct illumination, the differences between the two dosimeters increase with increasing room position. While this represents a somewhat extreme case, it is not an uncommon posture in office settings. For the other two postures, i.e., standing and seated, screen-facing, the diffuse lighting scenario produced minimal deviations between the eye-level dosimeter with FOV occlusion and the chest-worn dosimeter. This can be attributed to the FOV occlusion obstructing the upper and lower portions of the visual field, leaving most of the vertical FOV, such as toward a window in front of the subject, unaffected. In contrast, under both top lighting scenarios, the FOV occlusion obstructs the line of sight to the ceiling emitter, substantially reducing the eye-level measurements and thereby increasing deviations from the chest-worn measurements. Excluding the seated, desk-facing posture under diffuse side lighting, the deviations relative to the occluded eye-level dosimeter range from −18.6% to 141.2% for the upper chest and from −24.0% to 117.0% for the lower chest. The largest deviations occur under directional top lighting because the direct (downward) component of the emitter is largely obstructed by the FOV occlusion. Chest-worn dosimeter placement. a Dosimeter positions on the chest (upper-central, lower-central, upper-lateral, and lower-lateral). b Subject’s view direction (D) and dosimeter (sensor) normals (n). c Rotational dosimeter displacement (RDD) at the four chest positions, considering all twelve subjects in three postures. Each dot represents one subject in one posture. The inset rendering shows an example `view' from a chest-worn dosimeter, capturing the desk and arms of the subject. Figure and caption reproduced from the preceding study40. As noted in the introduction of this article, real-world studies investigating the in-situ performance of chest-worn dosimeters for measuring eye-level light exposure were typically conducted under naturalistic conditions, meaning that measurements were taken under various illumination fields and subject postures. While limitations in the intrinsic photometric sensor performance of real dosimeters may distort the reported results in these studies, we can compare our results to identify if measurements under real illumination fields are generally within our theoretical, i.e., simulation-based, range of deviations. Two studies, both with just a single participant, reported measurement deviations between chest-worn and eye-level dosimeters under solely indoor illumination fields. The first36 reported that the median luminous exposure at the chest was approximately 2% lower than at the eyes. The second35 found no significant differences between melanopic EDI (log lux) measured at the chest and eyes. In that study, the mean melanopic EDI was higher at the chest than at the eyes in three out of eight indoor scenarios (here, a scenario refers to a certain illumination field and subject posture). Thus, the findings of both studies fall within the range of measurement deviations observed in our results. When also considering studies conducted under both indoor and outdoor illumination fields, we observe conflicting claims, some of which align with our results. A recent study34 reported that eye-level illuminances were significantly lower than chest-level measurements, consistent with the larger range of possible positive deviations observed in our study. Contrarily, a single-participant study38 found that melanopic EDI measured on the chest was approximately half of that at the eyes, suggesting substantial underprediction, outside the range of our deviations. In contrast to both studies, an earlier study29 reported chest- and eye-level measurements being similar with no significant differences, which falls within our range of deviations. In our results, chest-worn dosimeters, when evaluated relative to an eye-level dosimeter with FOV occlusion, generally overestimated corneal light exposure. This finding is consistent with a recent study45 that investigated FOV occlusion across three ceiling-based electric lighting conditions in an office environment, and found a decrease of approximately 29% in eye-level melanopic EDI due to FOV occlusion. Similarly, a study with a single subject working in an office comparing dosimeter measurements with a hemispherical FOV and FOV occlusion reported a decrease in eye-level melanopic EDI of 22%46. While both studies focused on eye-level measurements only, i.e., not comparing to chest-worn dosimeter measurements, they support the observation that eye-level light exposure is substantially reduced when FOV occlusion is considered under top lighting. In summary, real-world studies tend to report results toward the lower end of the range of measurement deviations observed in our results. It should be noted, however, that when measurements are taken over time under varying illumination fields, both under- and overprediction may occur, and the averaged deviations will therefore tend to approach zero. In our simulations, the largest deviations were observed for the seated, desk-facing posture. This does not imply that such deviations cannot occur in practice, but rather that they may be less frequently observed under real-world conditions. From the previous, it is abundantly clear that deviations between chest-worn and eye-level measurements are highly dependent on the specific illumination field, as well as the subject’s posture. As such, it is not possible to define a single, fixed in-situ measurement uncertainty for chest-worn dosimeters. However, it is important to recognize that this uncertainty, i.e., the possible deviation from actual eye-level light exposure, can be substantial. Rather than aiming to generalize a fixed in-situ error value, it is more valuable to understand the underlying causes of these deviations and identify strategies to mitigate them. To do so, we must understand the contributions of TDD, RDD, and BSO factors to the total (mean) deviations between chest-worn and eye-level measurements. The contributions of TDD, RDD, and BSO factors to the mean relative differences between chest-worn and eye-level measurements are presented in Fig. 4. The contribution of the TDD is generally negative, up to 50% (although typically within 25%). In both top lighting scenarios, a chest-worn dosimeter is positioned farther from the emitter than the eyes. As radiant intensity decreases with distance, this spatial displacement results in lower measurements at the chest than at the eyes, hence the negative TDD contribution. Under diffuse side lighting, the contribution of TDD is less pronounced, as the distances from the emitter to the chest and to the eyes are similar. Regarding the RDD, for nearly all conditions, it accounts for the largest (positive) contribution to the mean relative differences. Across the four chest positions, RDD angles frequently exceed 30°, and are typically oriented upward (Fig. 6), meaning that the chest-worn dosimeter’s normal is angled more toward the ceiling than the subject’s view direction. Under both top lighting scenarios, this upward orientation results in the chest-worn dosimeter recording more direct illumination than the eye-level dosimeter, leading to the positive bias. Similarly, under diffuse side lighting, when subjects look down at the desk, the chest-worn dosimeters are oriented more upward, toward the emitter, again resulting in positive bias. In contrast, under the same diffuse side lighting, when the subject’s view direction is directly toward the emitter (in the standing and seated, screen-facing posture), the RDD contribution becomes slightly negative, since the chest-worn dosimeters are oriented slightly away from the emitter, as visible in the fisheye images in Fig. 9c. BSO consistently contributes negatively to the mean relative differences, which aligns with expectations given that the reference eye-level position is largely free of occlusion by the body. BSO is most pronounced under directional top lighting, where occlusions near the edges of the dosimeter’s FOV substantially block the direct illumination (Fig. 8c), as visible in the fisheye images in Fig. 9b. Additionally, the contribution of BSO is especially large for seated subjects looking down at the desk. In these cases, the subject’s arms and face occlude the dosimeter’s FOV, as can be seen in the visualizations of BSO in the preceding study40. These findings suggest that the primary consideration when placing a dosimeter on the chest for collecting (eye-level) PLE data should be its orientation relative to the subject’s view direction (RDD). If the contribution of this factor is minimized, though still resulting in a slight positive difference, the negative contributions of TDD and BSO may offset it. Example renderings of dosimeter views. The dosimeters are worn by a standing subject at the 3 m room position under a diffuse top lighting, b directional top lighting, and c diffuse side lighting. The eye-level view with human FOV occlusion is partially shaded by (non-reflecting) geometry just in front of the sensor. The renderings were made using the Radiance rpict program. Room schematic and three lighting scenarios. a Positions at which the subject models were placed in the room. Subject models were simulated in three postures; only the standing posture is illustrated. Simulations were conducted under b diffuse top lighting, c directional top lighting, and d diffuse side lighting. Each lighting scenario is flanked by renderings, generated using the Radiance rpict program, showing the view of the radiant environment from the 1 m and 7 m positions. The renderings have been normalized to a maximum radiance of 1, as shown in the horizontal legend at the base of the figure. In addition to the mean in-situ performance of chest-worn dosimeters across subjects, it is also important to consider inter-individual variability (Fig. 5), as light dosimetry is applied at the individual level. The range of variability appears to be strongly influenced by the subject’s posture. When considering an eye-level dosimeter without FOV occlusion, for standing and seated, screen-facing subjects, variability is relatively limited. This is likely due to the alignment of the upper torso with the view direction. In such cases, variability mainly results from differences in body morphology and clothing. In contrast, subjects in the seated, desk-facing posture show considerably larger variability. This can be attributed to greater dispersion in how subjects angled their torsos when writing, leading to greater variation in RDD, as reflected by the wider spread in Fig. 6. In absolute terms, the variability between our twelve subjects can be substantial. For example, in the seated, desk-facing posture, the interquartile range of the boxplots spans at a minimum 40.0%. This means that in our sample of twelve subjects, between two groups of three subjects, there is at least a 40% difference in deviations (mean relative differences between chest-worn and eye-level measurements). Additionally, for almost all considered lighting scenarios and postures, the inter-individual variability increased when FOV occlusion was applied to the eye-level dosimeter. This outcome was expected, as larger deviations inherently yield a larger range of individual (mean) relative differences. Note that, given our small sample size, the presented inter-individual variability may not be generalizable. While the desk-facing posture under which substantial inter-individual variability is observed may not be maintained throughout an entire measurement period, it highlights how the posture of a subject, even when performing the same task, can introduce uncertainty in PLE measurements. While variations in posture are part of natural behavior and do influence the actual eye-level light exposure, the range presented here reflects only the variability in deviations introduced by the use of chest-worn dosimeters. To further investigate the possible inter-individual variability resulting solely from differences in body morphology, we applied part of the methodology from this study to a larger dataset comprising approximately three thousand body scans of standing individuals in a separate study47. In that study, we simulated chest-worn and eye-level dosimeter measurements (without FOV occlusion) under the diffuse top and diffuse side lighting scenarios. Unclothed subjects were placed at the 3 m room position. Based on the leveraged dataset, dosimeter measurements on the upper chest, relative to the eyes, were found to be within a 20% spread for half of the subjects (interquartile range), with the full range extending to 79%. This range is slightly larger than that observed for our standing subjects under both lighting scenarios (Fig. 5), which may be expected, as a larger subject sample is more likely to include more extreme cases. The observed range of possible deviations between chest-worn and eye-level measurements (Fig. 3), along with the substantial inter-individual variability (Fig. 5), should raise concern regarding the reliability of chest-worn dosimeters for quantifying eye-level PLE in light-dosimetry field studies. As demonstrated through the disaggregation of TDD, RDD, and BSO factors (Fig. 4), the primary contributor to these discrepancies is the RDD of the chest-worn dosimeter. To minimize deviations between chest-worn dosimeter measurements and actual eye-level light exposure, it is worthwhile to consider an individualized, though simple, approach. For each subject, a chest position with minimal RDD should be selected. It is acceptable for this position to be slightly lower on the chest, as TDD appears to have a smaller influence on the results. While BSO is generally not a major concern, positions with high BSO potential should be avoided. Selecting a chest position with minimal RDD is relatively straightforward if the subject’s typical activities are known. For example, in the case of desk work, a chest position with an orientation approximating the view direction (minimal RDD) under seated, forward-facing postures should be selected. The false-color plots of RDD, as well as TDD and BSO, across the upper bodies of our subjects presented in the preceding study40, provide a visual means for identifying such positions. In situations where subjects are expected to consistently engage in activities that result in misalignment between their upper torso and view direction, such as writing at a desk, or when a wide range of postures is anticipated during the measurement period of interest, a large RDD may occur. In such cases, it is advisable to consider a head-mounted dosimeter. If an individualized approach is not feasible, it is important to recognize that allowing participants to freely choose the dosimeter position on the chest may introduce considerable uncertainty into the PLE data. Alternatively, a standardized chest position can be used consistently across all subjects. In our results, dosimeters placed on the lower chest had the smallest range of possible differences (−21.2% to 63.8% for the lower-central and −26.2% to 57.5% for the lower-lateral, Fig. 3), located approximately 0.35 m below the eyes (Fig. 6). However, even with standardized placement, substantial inter-individual variability may still arise due to differences in how the subjects posture themselves, as well as their body morphology and clothing (Fig. 5). When conducting measurements in indoor environments with predominantly top lighting, it is important to recognize that in our results a standard chest-worn dosimeter, i.e., without a FOV occlusion mask, typically recorded light levels that were at least about 25% higher than those actually experienced by the eyes (right sides of the panels in Fig. 3). This overprediction results from two factors: the absence of a FOV occlusion mask and the general tendency of chest-worn dosimeters to measure higher light levels than eye-level dosimeters under these lighting scenarios, largely due to the RDD. Hence, when using chest-worn dosimeters to estimate corneal light exposure, it is strongly recommended to either use a dosimeter that incorporates some shading device to represent human FOV occlusion45,46 or to interpret the collected data with appropriate caution and critical evaluation. While minimizing measurement discrepancies is relevant in most contexts, it’s important to note that, for any study employing light dosimetry, the required accuracy of the obtained PLE data depends inherently on the intended application. For instance, PLE data are used to determine the duration during which an individual’s PLE exceeds a given threshold, with thresholds based on PLE recommendations in the context of NIF effects of light48. In environments with high daytime light levels, even when a chest-worn dosimeter substantially underpredicts actual eye-level light exposure, thresholds may still be met simply because of the abundance of available light. In contrast, as evening light exposure is a key modulator of NIF effects49–51, as reflected by the low recommended evening PLE levels48, underprediction of the chest-worn dosimeter may result in critical misinterpretation of the PLE data. This study employed a novel hybrid measurement-simulation approach to overcome limitations of real-world methodologies. The results provide insight into the measurement behavior of such dosimeters under controlled yet realistic conditions. Nonetheless, because simulations almost always represent markedly simplified models of reality, certain considerations arise from such an approach. First, the accuracy of the simulated light interactions in the modeled indoor environment must be considered. The model outputs obtained using the Radiance parameters in Table 2 were unambiguous as reliable convergence was achieved. Therefore, a (real-world) validation experiment was considered superfluous in addition to being beyond the scope of the current study. Nonetheless, the simulations were based on a ‘gray world’. While the Radiance rtrace program standard outputs irradiances across three channels (RGB), all emitters and material properties were assigned equal values across these channels, thereby neglecting the spectral composition. In contrast, real-world emitters have distinct spectral power distributions, and materials have wavelength-dependent reflectances. Hence, our results pertain only to the geometric factors influencing illumination. When a dosimeter is used to measure quantities such as melanopic EDI, additional differences between chest-worn and eye-level measurements may be introduced. For example, if a chest-worn dosimeter primarily measures light from ceiling-mounted luminaires, while the eyes are mainly exposed to light from a window or computer screen, the spectral content of each source may differ, leading to additional differences between both measurement positions beyond those revealed in our simulations. To illustrate, for seated individuals in offices, the spectral composition of screen illumination can have a substantial impact on the eye-level light exposure when considering melanopic EDI52. Consequently, in such environments, the deviations observed under the diffuse side lighting scenario, which is a generalized representation of illumination from a screen in an otherwise dark room, may be quite prevalent. Table 2Radiance rtrace parameters used to simulate dosimeter measurementsdsdcdtabaaadaras0.0001106020485121024 Radiance rtrace parameters used to simulate dosimeter measurements Second, the design choices for the lighting scenarios warrant consideration. The scenarios used in this study represent simplified indoor environments. As such, the resulting illumination fields are idealized, whereas real-world lighting conditions are typically more heterogeneous. Greater heterogeneity is likely to result in larger differences between chest-worn and eye-level measurements than those observed under our scenarios. Additionally, only three lighting scenarios were considered in the current study. Therefore, for future work, it would be relevant to consider a variety of heterogeneous, real-world illumination scenarios to further investigate the performance of chest-worn dosimeters. Third, the positioning of the dosimeters and the simulation of their measurements introduce uncertainty. Body-worn dosimeters were oriented according to the surface normal of the subject model at the respective location. Since only twelve participants were used to generate the subject models, they serve primarily as illustrations of different body morphologies. In addition, slight differences in actual dosimeter orientation may be introduced due to droop, the tendency of a wearable device to tilt due to gravity, depending on how it is attached to clothing. Moreover, the view directions of the subjects were also somewhat simplified. For the seated, screen-facing subjects, the view direction was along the horizontal axis. Typically, seated subjects looking at a screen have a slight downward view direction45,53, which may result in a slightly larger RDD than that presented here. Additionally, we used theoretical (simulated) dosimeters. Real dosimeters introduce uncertainty in the measurements due to their intrinsic photometric sensor performance. This intrinsic performance can be assessed using established guidelines that describe various quality indices characterizing the sensor’s performance54–56. While older dosimeters often scored rather poorly on some of these indices (e.g., see refs. 29,36,39,57,58), more recent devices show improved performance33,35,59, and calibration and predictive models can be leveraged to further enhance performance60. Though the performance of wearable dosimeters remains inferior to that of conventional (non-miniaturized) laboratory photometers. In particular, a poor angular response can lead to inaccurate readings when light arrives at oblique angles relative to the sensor normal, effectively introducing a form of FOV occlusion, to some extent, similar to that caused by the human face. Therefore, when using actual dosimeters, the observed deviations due to the RDD and BSO factors may be compounded or mitigated by the device’s own limitations. Such an effect could be investigated by characterizing the angular sensitivity of real dosimeters across their FOV and incorporating this sensitivity into simulations to evaluate its effect on recorded light exposure. Lastly, the modeling of human FOV occlusion affects corneal light exposure simulations. The occlusion geometry modeled in front of the eye-level dosimeter was based on a generalized representation of the shading of the human eye by facial features as defined in CIE S 026:201844, rather than individualized geometry derived from our 3D body scans. As this study considered generalized lighting scenarios, adopting a standardized FOV occlusion was deemed most appropriate, as it facilitates comparison with other studies. We note, however, that this definition is a simplification and may not be representative of all facial morphologies, including those of our participants. Moreover, additional inter-individual differences are introduced when considering the fraction of corneal light exposure that reaches the retina, as this is modulated by the gaze direction of the pupils, pupillary area, and age-dependent eye characteristics61,62. Since the sensors of simulated dosimeters are represented as a point, the occlusion properties are invariant to the physical scale of the occlusion geometry. Thus, the geometry can be scaled to, effectively, an arbitrarily small size to allow close-proximity positioning with the 3D subject model whilst avoiding any intersections with the subject’s geometry. The actual physical size of the occlusion geometry does become an important factor when, instead of a point, the sensor is an area receptor – as is the case with all actual dosimeters and, indeed, all devices that measure light. This becomes an important consideration when occluding shapes, intended to represent the FOV as defined by the CIE, are added to measurement instruments and/or dosimeters to represent human FOV occlusion45,46. In short, the CIE description of a physical occluder can only be truly exact with regard to a single (sensor) point. Whereas an area receptor will always be subject, to some degree, to partial occlusion resulting in the effective ‘blurring’ of the occluding properties from the exact CIE description. The degree to which the performance of any particular combination of area receptor and physical occluder departs from the CIE description will depend on their relative sizes. The detailed modeling of such near-field occlusion effects for dosimeters is investigated by the authors in a separate study, which is nearing completion. To conclude, the results show that the chest-worn dosimeter’s performance is highly dependent on both the illumination field under which the measurements were conducted and the subject’s posture. Across the considered scenarios, chest-worn dosimeters generally overestimated light levels at the eye, with the smallest deviations (between −26.2% and 63.8%) for dosimeters worn on the lower chest. Substantial inter-individual variability was observed in the results. Using chest-worn dosimeters to estimate corneal light exposure, i.e., considering the occluded FOV of the eyes, under overhead lighting is problematic, with typical measurement deviations exceeding at minimum 25%. To reduce measurement discrepancies relative to the eye, as well as inter-individual variability, minimizing the RDD is key. Therefore, for light-dosimetry field studies, an individualized approach is recommended, in which, for each subject, a chest position is selected where the dosimeter’s orientation most closely aligns with the typical view direction during expected activities (postures). If such an approach is not feasible, then it is advisable to place the dosimeter on the lower chest, at approximately 0.35 m below the eyes. Methods Highly detailed 3D body scans of twelve participants in three postures were imported into three ‘virtual’ lighting scenarios for evaluation. The three postures used were: (1) standing, (2) seated at a desk while looking at a screen (screen-facing), and (3) seated while writing (desk-facing). Six participants identified as male and six as female. Their mean age was 33 ± 12 years, with a mean body mass index of 23.7 ± 2.5 kg/m2±. Details on the participants, scanning procedure, and conversion of the scans to subject models ready for lighting simulations are described in the preceding study40. The study was approved by the Ethical Review Board of the Eindhoven University of Technology (reference ERB2024BE76), and all participants provided informed consent. To quantify how the TDD, RDD, and BSO factors impact body-worn dosimeter measurements, we compared measurements of (simulated) dosimeters at the eye and on the body of these subject models under each of the three lighting scenarios. All simulations were conducted using Radiance v6.0a63. Dosimeter positions on the body To simulate dosimeter measurements on the body and at the eyes of each subject model, virtual sensors were defined in Radiance. For simplicity, we refer to these virtual sensors as dosimeters. For the body-worn dosimeter, we focus solely on devices worn on the chest. Although wrist-worn dosimeters are largely unobtrusive for the subject, they can have limitations in accurately capturing eye-level light exposure since their orientation (RDD) can be vastly different from the subject’s view direction36, and their field of view (FOV) may be occluded (BSO) or even completely covered by a shirt cuff. Accordingly, we believe they should be regarded only as a crude proxy measure for eye-level light exposure. A dosimeter near the eyes, typically integrated into spectacle frames, offers clear advantages in terms of accuracy for capturing eye-level light exposure. However, even modern devices remain relatively bulky, which can compromise the subject’s comfort. Additionally, the device’s visual appearance may influence the subject’s willingness to wear it under various (social) contexts64–66. Hence, balancing measurement performance and wear comfort, chest-worn dosimeters can be expected to offer the most practical compromise. Chest-worn dosimeters are often attached using a clip or a pendant, and consequently, a range of positions can be considered ‘chest-worn’. As it is not feasible to evaluate every possible position, we focused on four commonly used positions: at the upper-central, lower-central, upper-lateral, and lower-lateral chest, arranged in a square configuration measuring 0.1 by 0.1 m (Fig. 6a). The upper-central position is 0.25 m below the center of the eyes. Assuming lateral symmetry, no additional dosimeters were placed on the right side of the chest. Positions below the lower-central and lower-lateral chest were not considered, as they would typically be beneath a desk or table when seated. To determine the orientation of each dosimeter, we made the working assumption that a body-worn dosimeter aligns with the body’s outer, clothed contour, given its small size and weight. Accordingly, each dosimeter was oriented by aligning its sensor normal, i.e., the center of the sensor’s FOV, with the local surface normal of the subject model at the respective location (Fig. 6b). The method to determine these positions and the corresponding body surface normals of each subject model is described in the preceding study40. The dosimeter orientations relative to the subject’s view direction, or RDD, are given in Fig. 6c. To determine (reference) light measurements at the eye, a dosimeter was centered between the eyes of each subject model, placed at 1 cm in front of the skin (Fig. 6a). The dosimeter’s sensor normal aligned with the subject’s view direction (Fig. 6b). This placement ensured that, apart from the tip of the nose, the subject’s face did not obstruct the dosimeter’s FOV. Typically, the FOV of a dosimeter considers a full hemisphere, as illuminance (or irradiance) is defined as the integral of the cosine projected hemispherical luminance (or radiance). Yet this may not accurately represent the actual lighting reaching the outermost surface (cornea) of the eye, as the human FOV is occluded by facial features such as the brow, nose, and eyelids9,67,68, referred to as FOV occlusion. We refer to eye-level light measurements that include such FOV occlusion as corneal light exposure. If the dosimeter has a standard hemispherical FOV, we use the generic term eye-level light exposure. Under indoor illumination fields, FOV occlusion can lead to substantially lower measurements compared to dosimeters with a hemispherical FOV45,46,68. However, although including FOV occlusion may improve the accuracy of estimates of the light reaching the cornea, the use of standardized metrics remains essential to ensure comparability across studies and dosimeters. Therefore, we considered two eye-level references: a standard dosimeter with a hemispherical FOV and a dosimeter with FOV occlusion representing the shading by human facial features. To simulate measurements with FOV occlusion, geometry was modeled directly in front of the eye-level dosimeter and assigned a reflectance of zero. A Radiance rendering of the occlusion geometry (shaded blue) is shown in Fig. 7a. The gray disc represents the plane of the dosimeter sensor, and the plus marks the calculation point used for the lighting simulations. The 3D form of the occlusion geometry was based directly on the definition given in CIE S 026:201844. The CIE definition is described in terms of an angular fisheye projection, centered on the view direction, which has horizontal and vertical view angles of 180°. There are two zones of occlusion, referred to as upper and lower, which are defined, respectively, by two ‘horizontal’ chords at the 50° and 70° marks from the view direction on the (dashed) vertical view line, Fig. 7b. The occlusion geometry is a facet model derived from an exact analytical derivation of the CIE graphical representation. For the geometry, both the upper and lower occluders were each comprised of 100 (four-vertex) facets, which is sufficient to very closely approximate the (infinitely thin facet) exact representation. Views of the occlusion geometry from the perspective of the sensor point are shown by a pair of Radiance renderings employing angular (Fig. 7c) and hemispherical (Fig. 7d) fisheye projections. The occlusion geometry is shaded blue, and the unobstructed view is gray. As can be seen from the superposed graticule, the angular fisheye rendering of the occluder (Fig. 7c) matches exactly the CIE representation (Fig. 7b) because the same projection is used. Geometry representing the human field of view occlusion. a The geometry partially occluding the dosimeter’s field of view (FOV) is shown in blue. The geometry was modeled using planar facets, as the theoretical derivation of the CIE S 026:201844 definition does not require thickness. The center of the sensor is indicated by a plus. b The simulated FOV of the sensor. The FOV of the sensor simulated using Radiance with c an angular (ANGU) and d a hemispherical (HEMI) projection. Lighting scenarios The primary rationale behind the design of the lighting scenarios was to provide the bare minimum of environmental complexity, which would nevertheless allow the authors to quantify and disaggregate the combined and individual effects of the TDD, RDD, and BSO factors on chest-worn dosimeter measurements as a proxy for that received at the eye. Firstly, we decided on a single room configuration: a rectangular shoebox space with dimensions 8 × 2 × 2.5 m, Fig. 8a. The polygons comprising the room were assigned diffuse (gray) reflectance values of: 0.7 for the ceiling; 0.5 for the walls; and 0.3 for the floor. Subject models were assigned a diffuse (gray) reflectance of 0.4 to approximate the reflection of clothing and skin. The evaluations were carried out with the subject always facing one direction along the length of the room. There were seven evenly-spaced evaluation points, referred to as room positions, between the 1 m and 7 m marks, indicating respectively the ‘front’ and ‘back’ of the room. The plot symbols used for the graphical presentation of the results are indicated above the schematic of the subject: a cross for the position nearest the front of the room and a full circle for the remaining six positions. As noted, our aim is the evaluation of the effect of the TDD, RDD, and BSO factors on body-worn dosimeter measurements relative to eye-level measurements. Since we do not, at this stage, consider the spectral nature of the radiant environment, for simplicity, we will only use the terms irradiance (W m−2) and radiance (W m−2 sr−1) when describing the lighting scenarios. Thus, all absolute values will be shown normalized, usually to 1 or 100. Furthermore, we necessarily restrict the evaluation to comparing the relative differences between chest-worn and eye-level dosimeters. To investigate the in-situ performance of chest-worn dosimeters, it is desirable to consider ‘typical’ indoor environments, as performance is inherently dependent on the illumination field around the subject. In keeping with the aforementioned need for minimum environmental complexity, we decided, therefore, to consider just two types of generalized lighting scenarios: illumination from above and illumination from the side. To create the light source, the entire ceiling or part of the wall at the front of the room was designated as an emitter. For the ceiling-based emitter, two different outputs were defined: a completely diffuse distribution, referred to as the diffuse top lighting scenario (Fig. 8b), and a directional distribution, referred to as the directional top lighting scenario (Fig. 8c). The diffuse scenario represents either rooflight illumination or diffuse electric ceiling lighting. The directional scenario represents ceiling-mounted non-domestic electric lighting (e.g., in offices and schools) designed to direct light predominantly downwards to optimize task illumination and minimize glare for nearby occupants. Although the properties of such luminaires may vary, we approximated a typical distribution by modeling a directional emitter with an angular spread limited to within 45° of the source normal. For the wall-based emitter, a perfectly diffuse distribution was defined, referred to as diffuse side lighting (Fig. 8d). This scenario serves as a generalized representation of vertically incident illumination, such as skylight from a (vertical) window or light from a computer screen. Daylight can also have a strong directional component when sunlight enters a room through clear glazing. However, the presence of sunlight and its relative contribution to total daylight illumination indoors depend on numerous factors such as time of day/year, instantaneous sky conditions, and window orientation. Thus, it is not possible to devise a single, generalized daylight illumination field that includes sunlight. Flanking each schematic in Fig. 8 is a false-color 180° hemispherical fisheye image of the scene brightness (radiance) from the unobstructed viewpoint of a standing subject at the 1 m and 7 m positions. Since the focus is on relative differences rather than absolute light levels, the output of each emitter was arbitrarily set to a radiance of 1 W m−2 sr−1 per RGB channel, giving a visible radiance of unity. A linear mapping is used for the false-color scale, with the legend shown at the bottom of Fig. 8. Superposed on the images are concentric circle dashed lines drawn at 10° increments from the view-normal direction. With this view projection, equal areas of equal radiance contribute equally to the irradiance received at the viewpoint. The plus at the center marks the view direction. The first row of images depicts the diffuse top lighting scenario, Fig. 8b. The intention here is to reveal the effect of (gradually) changing position along the length of the room, and therefore a (gradual) change of the contribution of direct irradiance on the dosimeter. Starting at the 7 m position, i.e., the back of the room, the emitting ceiling in the rendering occupies almost a quarter of the hemispherical view. The walls and floor have a radiance of around 0.3 or less. At the 1 m position, the view now is mainly of the facing wall, with the emitter visible mainly in the periphery. The surface of the (diffuse) emitter has a uniform radiance of 1 from all viewpoints. For the second scenario, directional top lighting, the emitter was given a predominantly downward output distribution, Fig. 8c. Starting again at the 7 m position, the rendering shows the (visible) ceiling emitter now with a maximum (normalized) output at the periphery of visibility, almost directly ‘overhead’. Because, toward this ‘overhead’ region, the line of vision is almost parallel with the maximum directional output of the luminaire (which is directly downwards). In the rendering, the direct radiance of the emitter diminishes toward the front of the room – because these rays sample lower values of the emitter’s directional output distribution. The far end of the emitter (visible around 10° up from the view center) has a radiance comparable with that of the walls and floor. Moving now to the rendering from the 1 m position, the radiance distribution (in the hemispherical projection) appears very similar to that from the 7 m position. Lastly, the images for the diffuse side lighting scenario are shown in Fig. 8d. The image from the 1 m position shows the emitter dominating the FOV. Since the emitter is perfectly diffuse, it has a (normalized) uniform radiance equal to the scale maximum and is shaded bright yellow. The horizontal extent of the emitter subtends an angle of approximately 90° from this room position. The ceiling, walls, and floor have a radiance of around 0.2 or less. The image from the 7 m position shows the emitter, greatly diminished in comparison, subtending an angle less than 20°. The eye-level irradiance at each room position is a combination of direct radiation from the emitter and room reflected radiation from the surfaces visible in each image. For each lighting scenario, the irradiance values that correspond to the 1 m and 7 m images are shown in Table 1. The irradiance values have been normalized so that, in each case, the sum of the direct component (from the emitter) and the reflected component (from the environment) equals 100. Table 1Eye-level irradiance values at the 1 m and 7 m room positions under three lighting scenarios1 m irradiance7 m irradianceLighting scenarioDirectDirect +DirectDirect +onlyreflectedonlyreflectedDiffuse top lighting46.210066.8100Directional top lighting58.610064.2100Diffuse side lighting87.510069.7100The direct component and the combined direct and reflected component are given separately. Irradiance values have been normalized so that the sum of the direct and reflected components equals 100. Eye-level irradiance values at the 1 m and 7 m room positions under three lighting scenarios The direct component and the combined direct and reflected component are given separately. Irradiance values have been normalized so that the sum of the direct and reflected components equals 100. Lighting simulations Reference eye-level dosimeter measurements were simulated using two approaches: a standard irradiance sensor (dosimeter) with a hemispherical FOV and a sensor with the occluder geometry, representing human FOV occlusion. Measurements by the four chest-worn dosimeters were simulated using only an irradiance sensor with a hemispherical FOV. Example views of the eye-level and chest-worn dosimeters of a subject at the 3 m room position are shown in Fig. 9. These views were created by rendering 180° hemispherical fisheye images of the scene brightness (radiance), with the view orientation corresponding to that of each dosimeter. The images were normalized such that the maximum radiance is equal across all images. A logarithmic false-color grayscale was applied to reveal the room and subject details; thus, the shown visual contrast is exaggerated. Simulations were conducted using the Radiance rtrace program. The applied simulation parameters were defined based on convergence testing and are listed in Table 2. The irradiances measured by the dosimeters at the eyes (with a hemispherical FOV and human FOV occlusion) and at the four chest positions were simulated for the twelve subjects in three different postures, at each of the seven positions in the room, under the three lighting scenarios, resulting in a total of 4536 unique simulations. The simulated measurements of the chest-worn dosimeters were compared with those of the eye-level dosimeter (with both FOVs) to calculate the mean relative differences both across and within subjects. The relative difference D between simultaneously measured irradiance E by chest-worn dosimeter c, i.e., at the upper-central, lower-central, upper-lateral, or lower-lateral chest, and a dosimeter at the eyes of subject s, with posture p, at room position r, under lighting scenario l, and FOV v was calculated by: Using these relative differences, the mean differences across all subjects S, for each posture, room position, lighting scenario, and FOV were determined: Next, the mean differences within each subject (averaged over all positions R in the room), for each subject, posture, lighting scenario, and FOV were determined: To identify the contribution of each of the three in-situ factors, i.e., the TDD, RDD, and BSO, to the mean relative differences across subjects, each factor was also simulated separately. To assess the impact of solely TDD, simulations were conducted without the subject models and with the orientations (sensor normals) of the dosimeters on the chest set to the orientation of the dosimeter at the eyes (i.e., the subject’s view direction). To identify the impact of solely the RDD, subject models were again removed, and the location of the dosimeter at the eyes was set equal to the location of each dosimeter on the chest. Lastly, the impact of BSO was simply determined by comparing simulations with and without the subject models. Glossary of terms Body self-occlusion (BSO): The percentage of the (hemispherical) view of a body-worn dosimeter that is obstructed by a part of the body. The degree of BSO can be quantified with respect to a uniform source of ambient illumination. See ‘Field of view occlusion’ for the specific case of the eye. In-situ performance: The effectiveness of a body-worn dosimeter in measuring real-world light exposure representative of that at the eye. Intrinsic photometric sensor performance: The degree to which a dosimeter accurately measures light levels under standardized, controlled conditions. Rotational dosimeter displacement (RDD): The angular displacement between a defined view direction (e.g., forward-facing along the y-axis) and the body-worn dosimeter’s sensor surface normal. Translational dosimeter displacement (TDD): The Euclidean distance from the defined eye point (i.e., the left eye, the right eye, or the mid-point between the eyes) and the point on the body representing the location of the dosimeter. Field of view (FOV) occlusion: The reduced visual field at the eye resulting from the shape of the head, e.g., brow above the eye. FOV occlusion is a specific case of body self-occlusion. Acknowledgements We are grateful to Prof.dr. H. kort: https://orcid.org/0000-0001-9735-1015 for her feedback on the manuscript. During the preparation of this article, S.W.V. utilized Grammarly to improve the readability of the text. The authors reviewed the final version in its entirety and take full responsibility for the content. This study received no funding. Author contributions S.W.V.: Conceptualization, Formal analysis, Investigation, Methodology, Software, Visualization, Writing - Original Draft, Writing - Review & Editing. J.M.: Conceptualization, Methodology, Software, Visualization, Writing - Original Draft, Writing - Review & Editing. J.D.: Conceptualization, Supervision, Writing - Review & Editing. Data availability The datasets generated and/or analyzed during the current study are not publicly available as they are simulation-specific and primarily illustrative, but are available from the corresponding author on reasonable request. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement The datasets generated and/or analyzed during the current study are not publicly available as they are simulation-specific and primarily illustrative, but are available from the corresponding author on reasonable request.