Total body weight estimation by 3D camera systems: potential high-tech solutions for emergency medicine applications? A scoping review ====================================================================================================================================== * Mike Wells * Lara Nicole Goldstein * Terran Wells * Niloufar Ghazi * Abhijit Pandya * Borifoje Furht * Gabriella Engstrom * Muhammad Tanveer Jan * Richard Shih ## ABSTRACT **Background** Weight estimation is required in adult patients when weight-based medication must be administered during emergency care, as measuring weight is often not possible. Inaccurate estimations may lead to inaccurate drug dosing, which may cause patient harm. High-tech 3D camera systems driven by artificial intelligence might be the solution to this problem. The aim of this review was to describe and evaluate the published literature on 3D camera weight estimation methods. **Methods** A systematic literature search was performed for articles that studied the use of 3D camera systems for weight estimation in adults. Data on the study characteristics, the quality of the studies, the 3D camera methods evaluated, and the accuracy of the systems were extracted and evaluated. **Results** A total of 14 studies were included, published from 2012 to 2024. Most studies used Microsoft Kinect cameras, with various analytical approaches to weight estimation. The 3D camera systems often achieved a P10 of 90% (90% of estimates within 10% of actual weight), with all systems exceeding a P10 of 78%. The studies highlighted a significant potential for 3D camera systems to be suitable for use in emergency care. **Conclusion** The 3D camera systems offer a promising method for weight estimation in emergency settings, potentially improving drug dosing accuracy and patient safety. Weight estimates were extremely accurate. Importantly, 3D camera systems possess characteristics that could make them very appropriate for use during emergency care. Future research should focus on developing and validating this methodology in larger studies with true external and clinical validation. ## INTRODUCTION ### Background During the resuscitative care of critically ill or injured patients, an estimation of their weight is required when weight-based drug therapy is required, and actual patient weight cannot be measured. Measuring weight with scales is not always feasible as it is time-consuming and requires patient cooperation as well as a medically stable patient [1]. If a stand-on scale cannot be used due to the patient’s clinical condition, some emergency departments use scales that are imbedded in patient stretchers. These are costly, not universally available, and unproven in terms of accuracy during emergency care [2, 3]. An estimation of weight is, therefore, often required. Numerous studies have described or evaluated different methods of weight estimation in adults. These methods include (1) estimates by patients themselves, by family members, and by healthcare professionals; (2) formulas based on anthropometric measurements, such as the Lorenz formula; (3) dual length- and habitus-based tapes, such as the PAWPER XL-MAC tape; and (4) high-tech methods, such as 3D camera systems. Of all these methods, 3D camera systems have shown the greatest potential for highly accurate, rapid, easy-to-use estimation of weight [4]. These methods use real-time 3D images and previously trained artificial intelligence algorithms to generate estimates of weight. Existing 3D camera weight estimation methods have used different approaches with different cameras, software, costs, ease-of-use, applicability, and accuracy. ### Importance Inaccurate weight estimations may lead to inaccurate drug doses. This may cause patient harm through ineffective treatment (underestimation of weight) or adverse drug effects (overestimation of weight) [5]. One weight estimation researcher has written: “It cannot be considered to be good medical practice to use a weight estimation system that is known to be inaccurate” [6]. In adults, most current methods of weight estimation are simply not accurate enough [4]. It is possible that 3D camera methods of weight estimation may offer a solution to these problems. In addition, simple and easy-to-use methods are of special interest in emergency medicine. This is because more complex methods may be less easy to use and more prone to errors during the high cognitive loads experienced in emergency care [7]. High-tech 3D camera systems engineered for usability in the setting of emergency care may also be able to address this problem. ### Goals of this investigation Our aim in this scoping review was to review the available literature in which a 3D camera system was used to estimate a patient’s weight, with a medical indication as the ultimate purpose. We aimed to describe the performance and accuracy of 3D camera weight estimation systems, the types of cameras used, and the analytical and software methods used in the weight estimation process. ## METHODS This scoping review was based on the PRISMA for Scoping Reviews guidelines (PRISMA-ScR) [8]. ### Literature search A literature search was conducted using MEDLINE, EMBASE, IEEE Xplore, and Google Scholar. Eligible studies published between January 2012 and April 2024 were identified using the search strategy shown in Supplementary Table 1. ### Eligibility criteria Studies were included for further evaluation if they were peer reviewed, full length, English language papers containing original data. Studies evaluating any form of 3D camera weight estimation methodology, and in any type of participants were eligible for inclusion if an accurate measured weight was used as the standard reference. Studies on weight estimation not relevant to a clinical or hospital setting were excluded (e.g., weight estimation for forensic or non-medical applications). ### Selection of studies The titles and abstracts of the articles identified by the database search were manually screened by two researchers independently (MW, NG). The full texts of the selected reviews were then obtained and assessed for eligibility. Any differences in opinion were resolved by discussion and consensus. ### Critical appraisal of individual sources of evidence (Grading of quality of studies) Every included study was graded for quality of evidence using a modified Newcastle-Ottawa scale (NOS), as has been described previously (see Supplementary Table 2) [9]. Studies were downgraded if significant methodological weaknesses were present, e.g., if data presentation was incomplete or if performance outcome data was not appropriately presented or analyzed. An assessment of selective non-reporting or under-reporting of results in the studies was included in the Newcastle-Ottawa scale. Each study could score a minimum of zero stars and a maximum of 10 stars on the modified NOS. On this scale, a study with score from 6 to 10 has high quality, 4 to 5 has a moderate risk of bias, and 0 to 3 a very high risk of bias. ### Data charting process (Data extraction) Data extraction was conducted by one researcher (TW) using a standardized electronic data extraction form and was independently confirmed by another researcher (MW) for accuracy. ### Data items The following data was extracted: basic study information (region of origin, study population, sample size), study participant characteristics, 3D camera used, analytic method or software used for the weight estimation process, key findings, and the data presented on the performance or accuracy of weight estimation. ### Data synthesis (Map of outcomes) The findings of this scoping review were synthesized by presenting a descriptive and quantitative summary of the study characteristics using frequencies with percentages. The studies were grouped by the types of 3D cameras used, as well as the overall analytic approach, and summarized according to weight estimation outcomes. In terms of the quantitative analysis, the main outcomes of interest were metrics representing the performance of the weight estimation system. These included mean error or mean percentage error, which represented the estimation bias; the root mean square error, the mean absolute error, the root mean square percentage error or the mean absolute percentage error, which quantified the estimation precision; and the percentage of weight estimations that fell within 10% (P10) as well as within 20% (P20) of measured weight, which denoted overall accuracy. We considered the measures of overall accuracy (P10 and P20) to be the best indicator of overall performance, as we have described previously [10]. If P10 data was not reported it was imputed, whenever possible, from other reported metrics (mean absolute percentage error or mean percentage error). ## RESULTS No significant deviations from the protocol were noted. The details of the numbers of sources of evidence screened, assessed for eligibility, and included in the review, with reasons for exclusions at each stage are shown in Figure 1. ![Figure 1](http://medrxiv.org/http://medrxiv.stage.highwire.org/content/medrxiv/early/2024/08/20/2024.08.14.24311987.1/F1.medium.gif) [Figure 1](http://medrxiv.org/content/early/2024/08/20/2024.08.14.24311987.1/F1) Figure 1 The Preferred Reporting Items for Scoping Reviews (PRISMA-ScR) flow chart for article identification and selection. ### Characteristics of the included studies A total of 14 studies were included in this scoping review. The details of the included studies, including the study methodologies, the hardware and software used, and the weight estimation approaches are shown in Table 1. Two thirds of the studies (9/14 (64%)) were from Europe (all but one from Germany), with three studies (21%) from the USA, and two studies (14%) from elsewhere (one from Indonesia and one from Chile). View this table: [Table 1](http://medrxiv.org/content/early/2024/08/20/2024.08.14.24311987.1/T1) Table 1 Details of the included studies [11–24]. Abbreviations: NHANES – National Health and Nutrition Examination Survey; BMI – Body Mass Index; MATLAB-MATrix LABoratory; PCL – Point Cloud Library; ANN – Artificial Neural Network; MAPE – Mean Absolute Percentage Error; RANSAC – RANdom SAmple Consensus; SVR – Support Vector Regression; LibSVM – library for support vector machines; MAE – Mean Absolute Error; MPE – Mean Percentage Error; LOA – Limits of Agreement; P10 – percentage of estimates within 10% of actual weight; LMS – Least Mean Squares; CNN – Convoluted Neural Network; DBSCAN – Density-Based Spatial Clustering of Applications with Noise; ADAM – Adaptive Moment Estimation. Most of the studies (8/14 (57%)) prospectively collected data for data analysis, while six (43%) studies used existing data to develop or evaluate new analytic approaches. Only 4/14 (29%) studies compared 3D camera methods against other methods of weight estimation. The main aim of the study was to evaluate potential methods to estimate weight for drug dosing purposes in 6/14 (43%) studies, for CT contrast and radiation dosing in 5/14 (36%) studies, for nutritional or body habitus assessment in 2/14 (14%) studies, and other reasons in 1/14 (7%) study. ### Risks of bias and limitations across studies The methodological quality of most of the studies was good. Most studies (11/14 (79%)) had a low risk of bias on the Newcastle-Ottawa scale, one study (7%) had moderate risk of bias and two studies (14%) had a high risk of bias. However, incomplete data reporting or incomplete statistical analysis (5/14 studies (36%)) were common. Only 3/14 (21%) studies presented any form of subgroup analyses and no study provided comprehensive subgroup analyses by sex and weight-status. In addition, 2/14 (14%) studies had a sample size of 100 or fewer participants, and only 3/14 (21%) studies had a sample size of greater than 300 participants. These findings are summarized in Figure 2. ![Figure 2](http://medrxiv.org/http://medrxiv.stage.highwire.org/content/medrxiv/early/2024/08/20/2024.08.14.24311987.1/F2.medium.gif) [Figure 2](http://medrxiv.org/content/early/2024/08/20/2024.08.14.24311987.1/F2) Figure 2 Risk of bias traffic light plot and summary plot based on the Newcastle-Ottawa score. There were no studies with missing information or critical risks. Eight studies (57%) employed some form of appropriate internal validation of the developed model: split sample analysis in four studies, and cross validation in four studies. No 3D camera weight estimation system had a true external validation process. ### Camera technology and hardware The original Microsoft Kinect 1 camera was used in 7/14 (50%) studies, and the Microsoft Kinect 2 camera was used in 3/14 (21%) studies. A Siemens FAST 3D camera, a Siemens myExam 3D camera, and an Orbbec Astra camera were used in one study each (7%). The type of 3D camera used was not reported in two studies (14%). ### Fundamental approach used in the weight estimation methodology Multiple differences approaches were used to process 3D images and obtain a weight estimate from the depth data (see Table 2). Deep learning methods were used in the image preprocessing phase in 4/14 (29%) studies, and in the weight estimation phase in 9/14 (64%) studies. Most methods (9/14 (64%)) required no user input to facilitate the weight estimate calculations, with the exceptions of the methods of Pfitzner and colleagues which required gender as a manual input, and the method of Benalcazar *et al* which required information on clothing and hairstyle. View this table: [Table 2](http://medrxiv.org/content/early/2024/08/20/2024.08.14.24311987.1/T2) Table 2 Analytical approach to total body weight estimation in adults. ### Accuracy of weight estimates Unfortunately, only half of the studies (7/14 (50%)) provided comprehensive data on the performance of the weight estimation systems. Accuracy data (P10 – the percentage of estimates within 10% of actual weight) could be imputed in four additional studies. The accuracy data for each study is shown in Figure 3. Every study for which data was available exceeded the minimum acceptable accuracy standard of P10 >70% [10, 25]. ![Figure 3](http://medrxiv.org/http://medrxiv.stage.highwire.org/content/medrxiv/early/2024/08/20/2024.08.14.24311987.1/F3.medium.gif) [Figure 3](http://medrxiv.org/content/early/2024/08/20/2024.08.14.24311987.1/F3) Figure 3 The accuracy data (P10 – percentage of estimates within 10% of actual weight) for each of the 3D camera weight estimation systems. The studies marked with an asterisk (*) identify studies for which P10 data was imputed. The red dashed line indicates the minimum acceptable performance threshold of P10 = 70%. In the four studies in which direct, paired comparisons were made against other weight estimation systems, the following findings were notable: firstly, the 3D camera systems were always more accurate than guesstimates by healthcare providers (four studies); secondly, the 3D camera systems were always less accurate than participant self-estimates of weight (three studies). Comparative data was not available from the studies in which the 3D cameras achieved exceptionally high accuracy results. ### The suitability of 3D camera weight estimation systems for emergency and critical care There were 4/14 (29%) studies conducted in an environment designed to simulate an Emergency Department setting, and 5/14 (36%) studies conducted in, or with data from, a radiological suite. However, no study evaluated an estimation method during the provision of actual or simulated emergency care. ## LIMITATIONS There were several limitations to this review. Firstly, papers in the non-medical literature are less well indexed and searchable than in the medical literature. It is therefore possible that some relevant studies were missed. Secondly, the studies were from a very narrow range of geographical locations, which could limit the generalizability of the findings. Thirdly, the small sample sizes, the variable data reporting and statistical analysis, especially of subgroups of BMI, limited any comparisons between different 3D camera weight estimation systems. The need to impute data was also a limitation. Furthermore, few studies included a sufficiently diverse sample of participants with different ages, ethnic groups, height and weight ranges, and weight status (e.g., underweight, healthy weight, overweight, obese). ## DISCUSSION The current understanding of 3D camera-based weight estimation in adults, including its potential role during emergency care, has significant gaps. For example, when faced with a critically ill or injured patient in need of urgent weight-based drug therapy, but without any recorded weight, could a 3D camera system be used for estimating their weight? The significance of our review lies in its exploration of the currently available information on this topic. Our aim was to offer information and guidance to clinicians and researchers in this matter of important patient safety. The importance of the topic lies in the imperative for accurate of drug dosing: both treatment failure from underdosing and adverse events from overdosing can be significant threats to life. We identified and reviewed all the published literature on 3D camera weight estimation methods that could potentially be used during emergency medical care of adult patients. While some methods were primarily intended for nutritional assessment, others were devised and intended to guide acute medical interventions (e.g., to guide dosage of thrombolytic therapy in patients with acute ischemic stroke). ### Quality of the studies Although a few studies had inadequate data reporting and statistical analysis, most of the studies were methodologically sound. This provided a good evidence basis from which to draw preliminary conclusions. The lack of true external validation studies was a significant limitation in the field of 3D weight estimation, however. ### Camera technology The studies in this review made use of several different types of 3D cameras: structured light systems (e.g., Microsoft Kinect 1, Orbbec Astra) and time-of-flight systems (e.g., Microsoft Kinect 2). These camera systems are relatively old and, in some cases, no longer manufactured. Newer cameras have native software to perform many image processing tasks automatically: intrinsic and extrinsic calibration to ensure accurate depth measurements and color-depth alignment; automatically correct lens distortion; generate 3D point clouds from depth data; convert depth images to 3D coordinates automatically; detect and track human skeletons in real-time; automatically detect and track objects or faces within the camera’s field of view; provide bounding boxes or other positional data for detected objects; apply noise reduction and smoothing filters to depth data; perform edge detection and other image processing tasks. These newer 3D cameras are, therefore, likely to be better than those already tested. Existing research relating to the cameras themselves has been sparse, and future work needs to evaluate the most appropriate hardware system for use for weight estimation and in potential clinical emergency medicine applications. Table 3 provides a description of the different types of 3D cameras. View this table: [Table 3](http://medrxiv.org/content/early/2024/08/20/2024.08.14.24311987.1/T3) Table 3 3D cameras used in the weight estimation systems. ### Analytical approach to weight estimation The analytical approaches to weight estimation have evolved significantly in successive studies over the last decade. The earliest system described the use of a 3D camera to obtain biometric data which could be used in an equation derived from an anthropometric dataset [11]. Subsequent studies used depth data to calculate total body volumes (and, later, segmental body volumes), which are converted to weight estimates using density constants [13, 14, 16, 17]. The most recent methodologies have used deep learning to match a digital twin from a library of trained images against the point cloud data of a captured 3D image [24, 26]. This is perhaps the most flexible method, with the highest potential for accuracy. The use of deep learning both in image processing and in the weight estimation process has substantially improved the accuracy of weight estimates. ### Accuracy of weight estimation by 3D camera systems The best metric for evaluating the global performance of a weight estimation system is the overall accuracy (P10 and/or P20) [10]. In this review, each of the 11 weight estimation systems for which P10 data was available exceeded the minimum required accuracy threshold for a weight estimation system (P10>70%), as has been described previously [9, 25]. In fact, the lowest P10 was just below 80%, and four systems had a P10>95%. Overall, this performance data is remarkably good. To put this in context, a recent meta-analysis of weight estimation systems in adults showed that only patient self-estimates of weight approached this degree of accuracy but were inconsistent across studies [4]. In addition, self-estimates of weight were often not able to be provided by the sickest patients. In studies conducted during actual emergency medical situations, the number of patients unable to provide a self-estimate may be as high as 70 to 85% [27]. The evidence is thus clear that methods of weight estimation that do not rely on self-estimates must always be available [4]. The data from this scoping review shows that 3D camera systems could potentially fulfil this role if their performance holds up in larger scale clinical studies. The limited subgroup data presented suggested that weight estimation accuracy may be maintained in patients with obesity, but that models might need to improve their performance in patients who are underweight. Other noteworthy factors were that accurate weight estimation was achieved with several different 3D cameras, as well as with different processing and analytical approaches. This strongly supports the validity of the underlying principles, and predictable biological associations between body size, shape, and body weight. In addition, accurate weight estimation was even possible when patients were clothed or covered with light blankets [28]. ### Appropriateness for use during ED or prehospital emergency care The appropriateness of 3D camera weight estimation systems for use during emergency care was not explicitly studied, although several of the studies specifically intended their systems to be used for this purpose [13, 14, 16, 17, 29]. There are several factors that make fully evolved 3D camera systems ideal for use during emergency medical care. Firstly, they are quick. A weight estimation can be calculated in less than one second, even with the use of deep learning systems in both the image preprocessing and the weight estimation algorithms [16]. Secondly, they are highly automated. The system can automatically select the optimum image to use for the processing (useful for when patients are moving or uncooperative). No user input is required for the weight estimation: sex and height, which have significant associations with weight, can be estimated using deep learning. Thirdly, the system can compensate for patient posture and patient movement. Irrespective of whether the patient is supine, prone, or lateral, an accurate weight estimate can be obtained. Finally, light clothing or coverings do not interfere with weight estimation, as 3D camera systems can “see beneath the covers” using deep learning digital twin-based analyses. ### Future directions This is an important and exciting field for future research. The 3D camera systems need to be studied in larger samples, including representative numbers of underweight and obese patients, as well as patients from diverse population groups, to ensure generalizability. These methodologies also need to be evaluated in clinical environments, including during emergency care. Likewise, the research needs to include children. Future innovations could also include the estimation of ideal body weight and lean body weight to allow for precision weight-based dosing for all patients. At present, establishing these weights is complex and requires additional measurements and calculations. They are thus not routinely employed by emergency physicians. ## CONCLUSIONS The weight estimation accuracy of 3D camera-based systems represents a significant advancement in the field of automated measurement and analysis. These systems utilize precise depth sensing and 3D modeling to capture the volume and dimensions of objects or individuals with high accuracy. By integrating advanced algorithms and machine learning techniques, 3D camera-based systems can convert depth data into reliable weight estimates. When properly optimized, 3D camera-based weight estimation can achieve accuracy comparable to traditional weighing methods, providing a non-contact, efficient, and versatile potential solution for use during emergency care. However, it was clear from this review that additional, high quality prospective research is urgently needed in this field, as a matter of prioritizing patient safety during emergency care. ## Data Availability All data produced in the present study are available upon reasonable request to the authors ## Footnotes * The author list was updated with the correct version.. * Received August 14, 2024. * Revision received August 20, 2024. * Accepted August 20, 2024. * © 2024, Posted by Cold Spring Harbor Laboratory The copyright holder for this pre-print is the author. All rights reserved. The material may not be redistributed, re-used or adapted without the author's permission. ## REFERENCES 1. [1].Cattermole GN, Wells M. Comparison of adult weight estimation methods for use during emergency medical care. J Am Coll Emerg Physicians Open. 2021;2(4):e12515. doi: 10.1002/emp2.12515 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1002/emp2.12515&link_type=DOI) 2. [2].Wells M, Goldstein LN, Cattermole G. Development and validation of a length- and habitus-based method of total body weight estimation in adults. Am J Emerg Med. 2022;53:44–53. doi: 10.1016/j.ajem.2021.12.053 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.ajem.2021.12.053&link_type=DOI) 3. [3].Gerl H, Miko A, Nelson M, Godaire L. Are In-Bed Electronic Weights Recorded in the Medical Record Accurate? Medsurg Nurs. 2016;25(3):177–81, 201. doi: [https://doi.org/none](https://doi.org/none). 4. [4].Wells M, Goldstein LN, Alter SM, Solano JJ, Engstrom G, Shih RD. The accuracy of total body weight estimation in adults-A systematic review and meta-analysis. Am J Emerg Med. 2024;76:123–135. doi: 10.1016/j.ajem.2023.11.037 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.ajem.2023.11.037&link_type=DOI) 5. [5].Bailey B, Gaunt M, Grissinger M. Update on medication errors associated with incorrect patient weights. Pa Patient Saf Advis. 2016;13(2):50–57. doi: 6. [6].Luscombe MD, Owens BD, Burke D. Weight estimation in paediatrics: a comparison of the APLS formula and the formula ‘Weight=3(age)+7’. Emerg Med J. 2011;28(7):590–3. doi: 10.1136/emj.2009.087288 [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6NzoiZW1lcm1lZCI7czo1OiJyZXNpZCI7czo4OiIyOC83LzU5MCI7czo0OiJhdG9tIjtzOjUyOiIvbWVkcnhpdi9lYXJseS8yMDI0LzA4LzIwLzIwMjQuMDguMTQuMjQzMTE5ODcuMS5hdG9tIjt9czo4OiJmcmFnbWVudCI7czowOiIiO30=) 7. [7].Luten R, Wears RL, Broselow J, Croskerry P, Joseph MM, Frush K. Managing the unique size-related issues of pediatric resuscitation: reducing cognitive load with resuscitation aids. Acad Emerg Med. 2002;9(8):840–7. doi: 10.1111/j.1553-2712.2002.tb02175.x [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1197/aemj.9.8.840&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=12153892&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F08%2F20%2F2024.08.14.24311987.1.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000177345000012&link_type=ISI) 8. [8]. Andrea C. Tricco, Erin Lillie, Wasifa Zarin. PRISMA Extension for Scoping Reviews (PRISMA-ScR): Checklist and Explanation. Annals of Internal Medicine. 2018;169(7):467–473. doi: 10.7326/m18-0850 %m 30178033 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.7326/M18-0850&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=30178033&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F08%2F20%2F2024.08.14.24311987.1.atom) 9. [9].Wells M, Goldstein LN, Bentley A. The accuracy of emergency weight estimation systems in children-a systematic review and meta-analysis. Int J Emerg Med. 2017;10(1):29. doi: 10.1186/s12245-017-0156-5 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1186/s12245-017-0156-5&link_type=DOI) 10. [10].Wells M, Goldstein L. Appropriate Statistical Analysis and Data Reporting for Weight Estimation Studies. Pediatr Emerg Care. 2023;39(1):62–63. doi: 10.1097/PEC.0000000000002862 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1097/PEC.0000000000002862&link_type=DOI) 11. [11].Velardo C, Dugelay J-L, Pleari M, Ariano P. Building the space scale or how to weigh a person with no gravity. 2012 IEEE International Conference on Emerging Signal Processing Applications. 2012:67–70. doi: 10.1109/ESPA.2012.6152447 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1109/ESPA.2012.6152447&link_type=DOI) 12. [12].Nguyen TV, Feng J, Yan S. Seeing human weight from a single RGB-D image. J Comput Sci Technol. 2014;29(5):777–784. doi: 10.1007/s11390-014-1467-0 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1007/s11390-014-1467-0&link_type=DOI) 13. [13].Pfitzner C, May S, Merkl C, Breuer L, Köhrmann M, Braun J, et al. Libra3d: body weight estimation for emergency patients in clinical environments with a 3d structured light sensor. IEEE international conference on robotics and automation (ICRA). 2015:2888–2893. doi: 10.1109/ICRA.2015.7139593 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1109/ICRA.2015.7139593&link_type=DOI) 14. [14].Pfitzner C, May S, Nüchter A. Neural network-based visual body weight estimation for drug dosage finding. Medical Imaging 2016: Image Processing. 2016. doi: 10.1117/12.2216042 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1117/12.2216042&link_type=DOI) 15. [15].Benalcazar D, Benalcazar D, Erazo A. Artificial neural networks and digital image processing: An approach for indirect weight measurement. 2017 IEEE Second Ecuador Technical Chapters Meeting (ETCM). 2017:1–6. doi: 10.1109/ETCM.2017.8247457 16. [16].Pfitzner C, May S, Nüchter A. Evaluation of features from RGB-D data for human body weight estimation. IFAC-PapersOnLine. 2017;50(1):10148–10153. doi: 10.1016/j.ifacol.2017.08.1761 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.ifacol.2017.08.1761&link_type=DOI) 17. [17].Pfitzner C, May S, Nuchter A. Body Weight Estimation for Dose-Finding and Health Monitoring of Lying, Standing and Walking Patients Based on RGB-D Data. Sensors (Basel). 2018;18(5). doi: 10.3390/s18051311 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.3390/s18051311&link_type=DOI) 18. [18].Bigalke A, Hansen L, Heinrich MP, editors. End-to-end learning of body weight prediction from point clouds with basis point sets. Bildverarbeitung für die Medizin 2021; 2021: Springer. 19. [19].Dane B, Singh V, Nazarian M, O’Donnell T, Liu S, Kapoor A, et al. Prediction of Patient Height and Weight With a 3-Dimensional Camera. J Comput Assist Tomogr. 2021;45(3):427–430. doi: 10.1097/RCT.0000000000001166 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1097/RCT.0000000000001166&link_type=DOI) 20. [20].Geissler F, Heiss R, Kopp M, Wiesmuller M, Saake M, Wuest W, et al. Personalized computed tomography-Automated estimation of height and weight of a simulated digital twin using a 3D camera and artificial intelligence. Rofo. 2021;193(4):437–445. doi: 10.1055/a-1253-8558 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1055/a-1253-8558&link_type=DOI) 21. [21].Mameli M, Paolanti M, Conci N, Tessaro F, Frontoni E, Zingaretti P. Weight estimation from an RGB-D camera in top-view configuration. 2020 25th International Conference on Pattern Recognition (ICPR). 2021:7715–7722. doi: 10.1109/icpr48806.2021.9412519 22. [22].Naufal A, Anam C, Widodo CE, Dougherty G. Automated Calculation of Height and Area of Human Body for Estimating Body Weight Using a Matlab-based Kinect Camera. Smart Science. 2021;10(1):68–75. doi: 10.1080/23080477.2021.1983940 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1080/23080477.2021.1983940&link_type=DOI) 23. [23].Tamersoy B, Pîrvan FA, Pai S, Kapoor A. Accurate and robust patient height and weight estimation in clinical imaging using a depth camera. Medical Image Computing and Computer Assisted Intervention – MICCAI 2023. 2023:337–346. doi: 10.1007/978-3-031-43987-2_33 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1007/978-3-031-43987-2_33&link_type=DOI) 24. [24].Shahzadi I, Tamersoy B, Frohwein LJ, Subramanian S, Moenninghoff C, Niehoff JH, et al. Automated Patient Registration in Magnetic Resonance Imaging Using Deep Learning-Based Height and Weight Estimation with 3D Camera: A Feasibility Study. Acad Radiol. 2024;31(7):2715–2724. doi: 10.1016/j.acra.2024.01.029 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.acra.2024.01.029&link_type=DOI) 25. [25].Stewart D. Accuracy of the Broselow tape for estimating paediatric weight in two Australian Emergency Departments: University of Sydney; 2009 [Accessed 18 January 2024, Master of Biostatistics]. Available from: [https://ses.library.usyd.edu.au//bitstream/2123/6265/1/Declan%20Stewart%20WPP%20final.pdf](https://ses.library.usyd.edu.au//bitstream/2123/6265/1/Declan%20Stewart%20WPP%20final.pdf) 26. [26].Tamersoy B, Pîrvan FA, Pai S, Kapoor A. Accurate and Robust Patient Height and Weight Estimation in Clinical Imaging Using a Depth Camera. Medical Image Computing and Computer Assisted Intervention – MICCAI 2023. Lecture Notes in Computer Science 2023. p. 337–346. 27. [27].Garcia-Pastor A, Diaz-Otero F, Funes-Molina C, Benito-Conde B, Grandes-Velasco S, Sobrino-Garcia P, et al. Tissue plasminogen activator for acute ischemic stroke: calculation of dose based on estimated patient weight can increase the risk of cerebral bleeding. J Thromb Thrombolysis. 2015;40(3):347–52. doi: 10.1007/s11239-015-1232-4 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1007/s11239-015-1232-4&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=25995103&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2024%2F08%2F20%2F2024.08.14.24311987.1.atom) 28. [28].Bigalke A, Hansen L, Diesel J, Heinrich MP. Seeing under the cover with a 3D U-Net: point cloud-based weight estimation of covered patients. Int J Comput Assist Radiol Surg. 2021;16(12):2079–2087. doi: 10.1007/s11548-021-02476-0 [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1007/s11548-021-02476-0&link_type=DOI) 29. [29].Pfitzner C. Visual Human Body Weight Estimation with Focus on Medical Applications: Julius-Maximilians-Universitat Wurzburg; 2019