Comparison of synergy extrapolation and static optimization for estimating multiple unmeasured muscle activations during walking.

BACKGROUND: Calibrated electromyography (EMG)-driven musculoskeletal models can provide insight into internal quantities (e.g., muscle forces) that are difficult or impossible to measure experimentally. However, the need for EMG data from all involved muscles presents a significant barrier to the widespread application of EMG-driven modeling methods. Synergy extrapolation (SynX) is a computational method that can estimate a single missing EMG signal with reasonable accuracy during the EMG-driven model calibration process, yet its performance in estimating a larger number of missing EMG signals remains unknown. METHODS: This study assessed the accuracy with which SynX can use eight measured EMG signals to estimate muscle activations and forces associated with eight missing EMG signals in the same leg during walking while simultaneously performing EMG-driven model calibration. Experimental gait data collected from two individuals post-stroke, including 16 channels of EMG data per leg, were used to calibrate an EMG-driven musculoskeletal model, providing "gold standard" muscle activations and forces for evaluation purposes. SynX was then used to predict the muscle activations and forces associated with the eight missing EMG signals while simultaneously calibrating EMG-driven model parameter values. Due to its widespread use, static optimization (SO) applied to a scaled generic musculoskeletal model was also utilized to estimate the same muscle activations and forces. Estimation accuracy for SynX and SO was evaluated using root mean square errors (RMSE) to quantify amplitude errors and correlation coefficient r values to quantify shape similarity, each calculated with respect to "gold standard" muscle activations and forces. RESULTS: On average, compared to SO, SynX with simultaneous model calibration produced significantly more accurate amplitude and shape estimates for unmeasured muscle activations (RMSE 0.08 vs. 0.15, r value 0.55 vs. 0.12) and forces (RMSE 101.3 N vs. 174.4 N, r value 0.53 vs. 0.07). SynX yielded calibrated Hill-type muscle-tendon model parameter values for all muscles and activation dynamics model parameter values for measured muscles that were similar to "gold standard" calibrated model parameter values. CONCLUSIONS: These findings suggest that SynX could make it possible to calibrate EMG-driven musculoskeletal models for all important lower-extremity muscles with as few as eight carefully chosen EMG signals and eventually contribute to the design of personalized rehabilitation and surgical interventions for mobility impairments.

Opportunities and challenges of using generative AI to personalize educational assessment.

In line with the positive effects of personalized learning, personalized assessments are expected to maximize learner motivation and engagement, allowing learners to show what they truly know and can do. Considering the advances in Generative Artificial Intelligence (GenAI), in this perspective article, we elaborate on the opportunities of integrating GenAI into personalized educational assessments to maximize learner engagement, performance, and access. We also draw attention to the challenges of integrating GenAI into personalized educational assessments regarding its potential risks to the assessment's core values of validity, reliability, and fairness. Finally, we discuss possible solutions and future directions.

Machine learning in oncological pharmacogenomics: advancing personalized chemotherapy.

This review analyzes the application of machine learning (ML) in oncological pharmacogenomics, focusing on customizing chemotherapy treatments. It explores how ML can analyze extensive genomic, proteomic, and other omics datasets to identify genetic patterns associated with drug responses. This, in turn, facilitates personalized therapies that are more effective and have fewer side effects. Recent studies have emphasized ML's revolutionary role of ML in personalized oncology treatment by identifying genetic variability and understanding cancer pharmacodynamics. Integrating ML with electronic health records and clinical data shows promise in refining chemotherapy recommendations by considering the complex influencing factors. Although standard chemotherapy depends on population-based doses and treatment regimens, customized techniques use genetic information to tailor treatments for specific patients, potentially enhancing efficacy and reducing adverse effects.However, challenges, such as model interpretability, data quality, transparency, ethical issues related to data privacy, and health disparities, remain. Machine learning has been used to transform oncological pharmacogenomics by enabling personalized chemotherapy treatments. This review highlights ML's potential of ML to enhance treatment effectiveness and minimize side effects through detailed genetic analysis. It also addresses ongoing challenges including improved model interpretability, data quality, and ethical considerations. The review concludes by emphasizing the importance of rigorous clinical trials and interdisciplinary collaboration in the ethical implementation of ML-driven personalized medicine, paving the way for improved outcomes in cancer patients and marking a new frontier in cancer treatment.

Personalized strategies for brachytherapy of cervix cancer.

Among most tailored approaches in radiation oncology, the development of brachytherapy for the treatment of cervical cancer patients has benefited from various technological innovations. The development of 3D image-guided treatments was the first step for treatment personalization. This breakthrough preceded practice homogenization and validation of predictive dose and volume parameters and prognostic factors. We review some of the most significant strategies that emerged from the ongoing research in order to increase personalization in uterovaginal brachytherapy. A better stratification based on patients and tumors characteristics may lead to better discriminate candidates for intensification or de-escalation strategies, in order to still improve patient outcome while minimizing the risk of treatment-related side effects.

Reinforcement Learning Interventions on Boundedly Rational Human Agents in Frictionful Tasks.

Many important behavior changes are frictionful; they require individuals to expend effort over a long period with little immediate gratification. Here, an artificial intelligence (AI) agent can provide personalized interventions to help individuals stick to their goals. In these settings, the AI agent must personalize rapidly (before the individual disengages) and interpretably, to help us understand the behavioral interventions. In this paper, we introduce Behavior Model Reinforcement Learning (BMRL), a framework in which an AI agent intervenes on the parameters of a Markov Decision Process (MDP) belonging to a boundedly rational human agent. Our formulation of the human decision-maker as a planning agent allows us to attribute undesirable human policies (ones that do not lead to the goal) to their maladapted MDP parameters, such as an extremely low discount factor. Furthermore, we propose a class of tractable human models that captures fundamental behaviors in frictionful tasks. Introducing a notion of MDP equivalence specific to BMRL, we theoretically and empirically show that AI planning with our human models can lead to helpful policies on a wide range of more complex, ground-truth humans.

From patient classification to optimized treatment in ART: the AMPLITUDE Delphi consensus.

Introduction: A Delphi consensus was performed to evaluate expert opinions on the management of key aspects of ovarian stimulation. Methods: A Scientific Committee developed eleven statements for patient profiles corresponding to predicted ovarian responses (low, normal, and high) based on antral follicle count (AFC) and anti-Müllerian hormone (AMH). The statements were distributed (online survey) to French and Belgian fertility specialists. Consensus was reached when ≥66.7% of participants agreed or disagreed. Results: Among 52 respondents, a consensus agreement was reached for each patient profile for personalizing the initial dose of gonadotropin, taking age, weight, body mass index, nature of the cycle, and the decision to perform a fresh transfer or a freeze-all strategy into consideration. The respondents preferred a fresh transfer for low and normal responders and a freeze-all strategy in case of high risk of hyperstimulation, newly diagnosed uterine or tubal pathology and premature progesterone elevation. A consensus was reached for 10-15 oocytes as optimal oocyte target from the first round of voting. The panel agreed to increase the gonadotropin dose in case of insufficient response and preferred a GnRH antagonist protocol for a subsequent cycle in case of excessive response. Finally, a consensual answer was obtained for using LH/hCG activity in case of hypogonadotropic hypogonadism, advanced age, inadequate response during first stimulation and suspected FSH receptor polymorphism. Discussion: The AMPLITUDE consensus supports the importance of optimizing the ovarian stimulation protocol for patients undergoing assisted reproductive technology treatment. Additional studies could complete these findings and guide fertility specialists in their daily practice to improve ovarian stimulation outcomes.

Patient Initiated Follow-Up (PIFU): how can rheumatology departments start to reap the benefits? A consensus document.

Patient Initiated Follow-Up (PIFU) is gaining momentum in the NHS, aiming to optimize outpatient care amidst rising service demands. PIFU is valuable in rheumatology, where the increasing demand for ongoing management exacerbates the patient backlog. Importantly, PIFU has demonstrated comparable safety and outcomes to traditional care in numerous studies. PIFU empowers patients, drives personalized care, increases efficiency, and has the potential to reduce waiting lists by allowing services to focus on new and acute cases. Effective PIFU implementation includes careful selection of patients, educating patients and healthcare staff, well defined operational guidelines, and robust remote monitoring. Digital solutions can enhance PIFU through patient education, active remote monitoring and streamlined escalation. Electronic Patient Reported Outcome Measures (ePROMs) provide a suitable and safe metric to monitor patients remotely. Given the potential benefits, outpatient departments should consider investing in PIFU as a solution to current healthcare delivery challenges and as a means for future proofing clinical systems against increasing service demands.

Personalized Model-Driven Interventions for Decisions From Experience.

Cognitive models that represent individuals provide many benefits for understanding the full range of human behavior. One way in which individual differences emerge is through differences in knowledge. In dynamic situations, where decisions are made from experience, models built upon a theory of experiential choice (instance-based learning theory; IBLT) can provide accurate predictions of individual human learning and adaptivity to changing environments. Here, we demonstrate how an instance-based learning (IBL) cognitive model, implemented in a cognitive architecture (Adaptive Control of Thought-Rational), can be used to model an individual's decisions in a cybersecurity defense task, accounting for both population average and individual variances. The same IBL model structure with identical architectural parameters generates the full range of human behavior through stochastic memory retrieval processes operating over and contributing to unique experiences. Recurrence quantification analyses allow us to look beyond average behavior between and within individuals to sequential patterns of trial-to-trial behavior. We show how model-tracing and knowledge-tracing techniques can be used to align the model to an individual in real time to drive adaptive and personalized signaling algorithms for a cybersecurity defense system. We also present a method for introspecting into the cognitive model to gain further insight into the cognitive salience of features factored into individual decisions. The combination of techniques provides a blueprint for personalized modeling of individuals. We discuss the results and implications of this adaptive and personalized method for cybersecurity defense and more generally for intelligent artifacts tailored to individual differences in domains such as human-machine teaming.

Empowering Mental Health Monitoring Using a Macro-Micro Personalization Framework for Multimodal-Multitask Learning: Descriptive Study.

BACKGROUND: The field of mental health technology presently has significant gaps that need addressing, particularly in the domain of daily monitoring and personalized assessments. Current noninvasive devices such as wristbands and smartphones are capable of collecting a wide range of data, which has not yet been fully used for mental health monitoring. OBJECTIVE: This study aims to introduce a novel dataset for personalized daily mental health monitoring and a new macro-micro framework. This framework is designed to use multimodal and multitask learning strategies for improved personalization and prediction of emotional states in individuals. METHODS: Data were collected from 298 individuals using wristbands and smartphones, capturing physiological signals, speech data, and self-annotated emotional states. The proposed framework combines macro-level emotion transformer embeddings with micro-level personalization layers specific to each user. It also introduces a Dynamic Restrained Uncertainty Weighting method to effectively integrate various data types for a balanced representation of emotional states. Several fusion techniques, personalization strategies, and multitask learning approaches were explored. RESULTS: The proposed framework was evaluated using the concordance correlation coefficient, resulting in a score of 0.503. This result demonstrates the framework's efficacy in predicting emotional states. CONCLUSIONS: The study concludes that the proposed multimodal and multitask learning framework, which leverages transformer-based techniques and dynamic task weighting strategies, is superior for the personalized monitoring of mental health. The study indicates the potential of transforming daily mental health monitoring into a more personalized app, opening up new avenues for technology-based mental health interventions.

Multibody dynamics-based musculoskeletal modeling for gait analysis: a systematic review.

Beyond qualitative assessment, gait analysis involves the quantitative evaluation of various parameters such as joint kinematics, spatiotemporal metrics, external forces, and muscle activation patterns and forces. Utilizing multibody dynamics-based musculoskeletal (MSK) modeling provides a time and cost-effective non-invasive tool for the prediction of internal joint and muscle forces. Recent advancements in the development of biofidelic MSK models have facilitated their integration into clinical decision-making processes, including quantitative diagnostics, functional assessment of prosthesis and implants, and devising data-driven gait rehabilitation protocols. Through an extensive search and meta-analysis of over 116 studies, this PRISMA-based systematic review provides a comprehensive overview of different existing multibody MSK modeling platforms, including generic templates, methods for personalization to individual subjects, and the solutions used to address statically indeterminate problems. Additionally, it summarizes post-processing techniques and the practical applications of MSK modeling tools. In the field of biomechanics, MSK modeling provides an indispensable tool for simulating and understanding human movement dynamics. However, limitations which remain elusive include the absence of MSK modeling templates based on female anatomy underscores the need for further advancements in this area.

[The future of diabetes treatment]. / L'avenir des traitements du diabète.

Numerous therapeutic advances in the management of diabetes have been made in recent years, leading to the 2023 update of the Société francophone du diabète's position paper. These new treatments, for both autoimmune type 1 and type 2 diabetes, will continue to develop, offering patients personalized care.

Personalizing radiotherapy with adaptive radiotherapy: Interest and challenges.

Adaptive radiotherapy (ART) is a recent development in radiotherapy technology and treatment personalization that allows treatment to be tailored to the daily anatomical changes of patients. While it was until recently only performed "offline", i.e. between two radiotherapy sessions, it is now possible during ART to perform a daily online adaptive process for a given patient. Therefore, ART allows a daily customization to ensure optimal coverage of the treatment target volumes with minimized margins, taking into account only the uncertainties related to the adaptive process itself. This optimization appears particularly relevant in case of daily variations in the positioning of the target volume or of the organs at risk (OAR) associated with a proximity of these volumes and a tenuous therapeutic index. ART aims to minimize severe acute and late toxicity and allows tumor dose escalation. These new achievements have been possible thanks to technological development, the contribution of new multimodal and onboard imaging modalities and the integration of artificial intelligence tools for the contouring, planning and delivery of radiation therapy. Online ART is currently available on two types of radiotherapy machines: MR-linear accelerators and recently CBCT-linear accelerators. We will first describe the benefits, advantages, constraints and limitations of each of these two modalities, as well as the online adaptive process itself. We will then evaluate the clinical situations for which online adaptive radiotherapy is particularly indicated on MR- and CBCT-linear accelerators. Finally, we will detail some challenges and possible solutions in the development of online ART in the coming years.

Personalizing the shoulder rhythm in a computational upper body model improves kinematic tracking in high range-of-motion arm movements.

Musculoskeletal models of the shoulder are needed to understand the mechanics of overhead motions. Existing models implementing the shoulder rhythm are generic and might not accurately represent an individual's scapular kinematics. We introduce a method to personalize the shoulder rhythm of a computational model of the upper body that defines the orientations of the clavicle and scapula based on glenohumeral joint angles. During five static calibration poses, we palpate and measure the orientation of the scapula. We explore the importance of representing shoulder elevation by introducing clavicle elevation as a degree of freedom that is independent of the glenohumeral angles. For ten subjects, we record the five calibration poses, ten additional static poses, and dynamic arm raises covering the participants' full range of motion in each body plane using optical motion capture. We examine the data using a dynamically-constrained inverse kinematics analysis. Shoulder rhythm personalization, independent clavicle elevation, and both in combination reduce the average upper body marker tracking error compared to the generic model in the static poses (26 mm to 17-20 mm) and in the dynamic trials (22 mm to 14-17 mm). Only personalization reduces the average scapula marker error (51 mm to 36-38 mm) and scapula axis-angle error (15° to 10°) compared with the palpated ground truth measurements in the static poses, and in the dynamic trials at instances that best match the static poses (53 mm to 37-40 mm, 15° to 9°). Our results show that personalizing upper body models improves kinematic tracking. We provide our experimental data, model, and methods to allow researchers to reproduce and build upon our results.

[Finite element analysis of anatomic plate fixation for proximal clavicular fractures].

OBJECTIVE: To explore establishment and finite element analysis of personalized proximal clavicular anatomical plate screw fixation model. METHODS: A 40-year-old male healthy volunteer was selected and the finite element analysis modules of 3D reconstruction software Mimics 15.01, Hypermesh 2019 and Abaqus 2020 were used. The finite element model of anatomic plate at the proximal clavicle was established, and a vertical load of 250 N was applied to the distal end of long axis of clavicle about 15 mm, then the overall structure, plate and screw displacement cloud image, Mises stress distribution were observed. RESULTS: The displacement distribution of the overall structure shows the maximum displacement was distributed on the distal clavicle. Under the four conditions of normal upper limb weight, longitudinal clavicle fracture, oblique fracture and shoulder impact violence during fall, longitudinal clavicle fracture and oblique fracture, the maximum displacement were 1.04 mm, 1.03 mm, 1.35 mm and 1.33 mm, respectively. The displacement cloud map of titanium alloy steel plate showed the largest displacement was distributed near the distal clavicular bone, and the maximum displacement were 0.89 mm, 0.88 mm, 1.10 mm and 1.09 mm, respectively. The displacement cloud map of titanium alloy screw showed the largest displacement was distributed at the root of the distal screw, and the maximum displacement were 0.88 mm, 0.87 mm, 1.08 mm and 1.06 mm, respectively. Mises stress distribution showed the maximum stress was mainly distributed on titanium alloy plates and screws, and the stress on the clavicle was very small. Mises stress distribution cloud showed the maximum Mises stress was distributed at the second row of screw holes near the clavicle, and the maximum Mises stress were 673.1, 678.1, 648.5, 654.4 MPa, respectively. The maximum stresses of titanium alloy screws were 414.5, 417.4, 415.8 and 419.7 MPa, respectively. CONCLUSION: The biomechanical changes of personalized proximal clavicular anatomical plates are demonstrated by using 3D finite element method to provide biomechanical data for personalized proximal clavicular anatomical plates.

Optimal control of combination immunotherapy for a virtual murine cohort in a glioblastoma-immune dynamics model.

The immune checkpoint inhibitor anti-PD-1, commonly used in cancer immunotherapy, has not been successful as a monotherapy for the highly aggressive brain cancer glioblastoma. However, when used in conjunction with a CC-chemokine receptor-2 (CCR2) antagonist, anti-PD-1 has shown efficacy in preclinical studies. In this paper, we aim to optimize treatment regimens for this combination immunotherapy using optimal control theory. We extend a treatment-free glioblastoma-immune dynamics ODE model to include interventions with anti-PD-1 and the CCR2 antagonist. An optimized regimen increases the survival of an average mouse from 32 days post-tumor implantation without treatment to 111 days with treatment. We scale this approach to a virtual murine cohort to evaluate mortality and quality of life concerns during treatment, and predict survival, tumor recurrence, or death after treatment. A parameter identifiability analysis identifies five parameters suitable for personalizing treatment within the virtual cohort. Sampling from these five practically identifiable parameters for the virtual murine cohort reveals that personalized, optimized regimens enhance survival: 84% of the virtual mice survive to day 100, compared to 60% survival in a previously studied experimental regimen. Subjects with high tumor growth rates and low T cell kill rates are identified as more likely to die during and after treatment due to their compromised immune systems and more aggressive tumors. Notably, the MDSC death rate emerges as a long-term predictor of either disease-free survival or death.

Personalization of Intravaginal rings by droplet deposition modeling based 3D printing technology.

Intravaginal rings (IVRs) are long-acting drug device systems designed for controlled drug release in the vagina. Commercially available IVRs employ a one-size-fits-all development approach, where all patients receive the same drug in similar doses and frequencies, allowing no space for dosage individualization for specific patients' needs. To allow flexibility for dosage individualization, this study explores the impact of infill-density on critical characteristics of personalized IVRs, manufactured using droplet deposition modeling three-dimensional (3D) printing technology. The model drug was dispersed on the surface of thermoplastic polyurethane pellets using an oil coating method. IVR infill-density ranged from 60 to 100 %. The compatibility of the drug and matrix was assessed using thermal and spectroscopic analyses. The IVRs were evaluated for weight, porosity, surface morphology, mechanical properties, and in vitro drug release. The results demonstrated high dimensional accuracy and uniformity of 3D-printed IVRs, indicating the robustness of the printing process. Increasing infill-density resulted in greater weight, storage modulus, Young's modulus, Shore hardness, and compression strength, while reducing the porosity of IVRs. All IVRs showed a controlled drug release pattern when tested under accelerated conditions of temperature for 25 days. Notably, greater infill-densities were associated with a decrease in the percentage of drug released. Overall, the study demonstrated that infill-density was an important parameter for personalizing the critical characteristics of the 3D-printed IVRs to fit individual patient needs.

Did we personalize? Assessing personalization by an online reinforcement learning algorithm using resampling.

There is a growing interest in using reinforcement learning (RL) to personalize sequences of treatments in digital health to support users in adopting healthier behaviors. Such sequential decision-making problems involve decisions about when to treat and how to treat based on the user's context (e.g., prior activity level, location, etc.). Online RL is a promising datadriven approach for this problem as it learns based on each user's historical responses and uses that knowledge to personalize these decisions. However, to decide whether the RL algorithm should be included in an "optimized" intervention for real-world deployment, we must assess the data evidence indicating that the RL algorithm is actually personalizing the treatments to its users. Due to the stochasticity in the RL algorithm, one may get a false impression that it is learning in certain states and using this learning to provide specific treatments. We use a working definition of personalization and introduce a resampling-based methodology for investigating whether the personalization exhibited by the RL algorithm is an artifact of the RL algorithm stochasticity. We illustrate our methodology with a case study by analyzing the data from a physical activity clinical trial called HeartSteps, which included the use of an online RL algorithm. We demonstrate how our approach enhances data-driven truth-in-advertising of algorithm personalization both across all users as well as within specific users in the study.

A new modular neuroprosthesis suitable for hybrid FES-robot applications and tailored assistance.

BACKGROUND: To overcome the application limitations of functional electrical stimulation (FES), such as fatigue or nonlinear muscle response, the combination of neuroprosthetic systems with robotic devices has been evaluated, resulting in hybrid systems that have promising potential. However, current technology shows a lack of flexibility to adapt to the needs of any application, context or individual. The main objective of this study is the development of a new modular neuroprosthetic system suitable for hybrid FES-robot applications to meet these needs. METHODS: In this study, we conducted an analysis of the requirements for developing hybrid FES-robot systems and reviewed existing literature on similar systems. Building upon these insights, we developed a novel modular neuroprosthetic system tailored for hybrid applications. The system was specifically adapted for gait assistance, and a technological personalization process based on clinical criteria was devised. This process was used to generate different system configurations adjusted to four individuals with spinal cord injury or stroke. The effect of each system configuration on gait kinematic metrics was analyzed by using repeated measures ANOVA or Friedman's test. RESULTS: A modular NP system has been developed that is distinguished by its flexibility, scalability and personalization capabilities. With excellent connection characteristics, it can be effectively integrated with robotic devices. Its 3D design facilitates fitting both as a stand-alone system and in combination with other robotic devices. In addition, it meets rigorous requirements for safe use by incorporating appropriate safety protocols, and features appropriate battery autonomy, weight and dimensions. Different technological configurations adapted to the needs of each patient were obtained, which demonstrated an impact on the kinematic gait pattern comparable to that of other devices reported in the literature. CONCLUSIONS: The system met the identified technical requirements, showcasing advancements compared to systems reported in the literature. In addition, it demonstrated its versatility and capacity to be combined with robotic devices forming hybrids, adapting well to the gait application. Moreover, the personalization procedure proved to be useful in obtaining various system configurations tailored to the diverse needs of individuals.

Predicting the outcome of psychotherapy for chronic depression by person-specific symptom networks.

Psychotherapies are efficacious in the treatment of depression, albeit only with a moderate effect size. It is hoped that personalization of treatment can lead to better outcomes. The network theory of psychopathology offers a novel approach suggesting that symptom interactions as displayed in person-specific symptom networks could guide treatment planning for an individual patient. In a sample of 254 patients with chronic depression treated with either disorder-specific or non-specific psychotherapy for 48 weeks, we investigated if person-specific symptom networks predicted observer-rated depression severity at the end of treatment and one and two years after treatment termination. Person-specific symptom networks were constructed based on a time-varying multilevel vector autoregressive model of patient-rated symptom data. We used statistical parameters that describe the structure of these person-specific networks to predict therapy outcome. First, we used symptom centrality measures as predictors. Second, we used a machine learning approach to select parameters that describe the strength of pairwise symptom associations. We found that information on person-specific symptom networks strongly improved the accuracy of the prediction of observer-rated depression severity at treatment termination compared to common covariates recorded at baseline. This was also shown for predicting observer-rated depression severity at one- and two-year follow-up. Pairwise symptom associations were better predictors than symptom centrality parameters for depression severity at the end of therapy and one year later. Replication and external validation of our findings, methodological developments, and work on possible ways of implementation are needed before person-specific networks can be reliably used in clinical practice. Nevertheless, our results indicate that the structure of person-specific symptom networks can provide valuable information for the personalization of treatment for chronic depression.

Robot-Mediated Nudges for Workplace Health: Not a One-Size-Fits-All Modeling Problem.

Prolonged sedentary behavior in the vast population of office and remote workers leads to increased cardiovascular and musculoskeletal health challenges, and existing solutions for encouraging breaks are either costly health coaches or notification systems that are easily ignored. A socially assistive robot (SAR) for promoting healthy workplace practices could provide the physical presence of a health coach along with the scalability of a notification system. To investigate the impact of such a system, we implemented a SAR as an alternative break-taking support solution and examined its impact on individual users' break-taking habits over relatively long-term deployments. We conducted an initial two-month-long study (N = 7) to begin to understand the robot's influence beyond the point of novelty, and we followed up with a week-long data collection (N = 14) to augment the dataset size. The resulting data was used to inform a robot behavior model and formulate possible methods of personalizing robot behaviors. We found that uninterrupted sitting time tended to decrease with our SAR intervention. During model formulation, we found participant responsiveness to the break-taking prompts could be classified into three archetypes and that archetype-specific adjustments to the general model led to improved system success. These results indicate that break-taking prompts are not a one-size-fits-all problem, and that even a small dataset can support model personalization for improving the success of assistive robotic systems.