Multimodal wearable EEG, EMG and accelerometry measurements improve the accuracy of tonic-clonic seizure detection.

OBJECTIVE: This paper aims to investigate the possibility of detecting tonic-clonic seizures (TCSs) with behind-the-ear, two-channel wearable electroencephalography (EEG), and to evaluate its added value to non-EEG modalities in TCS detection. METHODS: We included 27 participants with a total of 44 TCSs from the European multicenter study SeizeIT2. The wearable Sensor Dot (SD; Byteflies) was used to measure behind-the-ear EEG, electromyography (EMG), electrocardiography (ECG), accelerometry (ACC) and gyroscope (GYR). We evaluated automatic unimodal detection of TCSs, using sensitivity, precision, false positive rate (FPR) and F1-score. Subsequently, we fused the different modalities and again assessed performance. Algorithm-labeled segments were then provided to two experts, who annotated true positive TCSs, and discarded false positives (FPs). RESULTS: Wearable EEG outperformed the other single modalities with a sensitivity of 100% and a FPR of 10.3/24h. The combination of wearable EEG and EMG proved most clinically useful, delivering a sensitivity of 97.7%, an FPR of 0.4/24h, a precision of 43%, and an F1-score of 59.7%. The highest overall performance was achieved through the fusion of wearable EEG, EMG, and ACC, yielding a sensitivity of 90.9%, an FPR of 0.1/24h, a precision of 75.5%, and an F1-score of 82.5%. CONCLUSIONS: In TCS detection with a wearable device, combining EEG with EMG, ACC or both resulted in a remarkable reduction of FPR, while retaining a high sensitivity. SIGNIFICANCE: Adding wearable EEG could further improve TCS detection, relative to extracerebral-based systems.

Avoiding Post-processing with Event-Based Detection in Biomedical Signals.

OBJECTIVE: Finding events of interest is a common task in biomedical signal processing. The detection of epileptic seizures and signal artefacts are two key examples. Epoch-based classification is the typical machine learning framework to detect such signal events because of the straightforward application of classical machine learning techniques. Usually, post-processing is required to achieve good performance and enforce temporal dependencies. Designing the right post-processing scheme to convert these classification outputs into events is a tedious, and labor-intensive element of this framework. METHODS: We propose an event-based modeling framework that directly works with events as learning targets, stepping away from ad-hoc post-processing schemes to turn model outputs into events. We illustrate the practical power of this framework on simulated data and real-world data, comparing it to epoch-based modeling approaches. RESULTS: We show that event-based modeling (without tailored post-processing) performs on par with or better than epoch-based modeling with extensive post-processing. CONCLUSION: These results show the power of treating events as direct learning targets, instead of using ad-hoc post-processing to obtain them, severely reducing design effort. Significance The event-based modeling framework can easily be applied to other event detection problems in signal processing, removing the need for intensive task-specific post-processing.

Learning Robust Representations of Tonic-Clonic Seizures with Cyclic Transformer.

Tonic-clonic seizures (TCSs) pose a significant risk for sudden unexpected death in epilepsy (SUDEP). Previous research has highlighted the potential of multimodal wearable seizure detection systems in accurately detecting TCSs through continuous monitoring, enabling timely alarms and potentially preventing SUDEP. However, such multimodal systems carry a higher risk of sensor malfunction. In this paper, we propose a cyclic transformer approach to address these challenges. The cyclic transformer learns a robust representation by performing circular modal translations between the source and target modalities. It leverages back-translation as regularization technique to enhance the discriminative power of the learned representation. Notably, the proposed cyclic transformer is trained on paired multimodal data but requires only a single source modality during deployment. This characteristic ensures the robustness of the cyclic transformer to perturbations or missing information in the target modality. Experimental results demonstrate that the proposed cyclic transformer achieves competitive performance compared with existing multimodal systems. While both approaches were trained using EEG and EMG data, the cyclic transformer exclusively employs EEG data for testing, diverging from the state-of-the-art's utilization of both EEG and EMG data during test. This showcases the effectiveness of the cyclic transformer in multimodal TCSs detection, offering a promising approach for enhancing the accuracy and robustness of seizure detection systems while mitigating the risks associated with sensor malfunction.

U-PASS: An uncertainty-guided deep learning pipeline for automated sleep staging.

With the increasing prevalence of machine learning in critical fields like healthcare, ensuring the safety and reliability of these systems is crucial. Estimating uncertainty plays a vital role in enhancing reliability by identifying areas of high and low confidence and reducing the risk of errors. This study introduces U-PASS, a specialized human-centered machine learning pipeline tailored for clinical applications, which effectively communicates uncertainty to clinical experts and collaborates with them to improve predictions. U-PASS incorporates uncertainty estimation at every stage of the process, including data acquisition, training, and model deployment. Training is divided into a supervised pre-training step and a semi-supervised recording-wise finetuning step. We apply U-PASS to the challenging task of sleep staging and demonstrate that it systematically improves performance at every stage. By optimizing the training dataset, actively seeking feedback from domain experts for informative samples, and deferring the most uncertain samples to experts, U-PASS achieves an impressive expert-level accuracy of 85% on a challenging clinical dataset of elderly sleep apnea patients. This represents a significant improvement over the starting point at 75% accuracy. The largest improvement gain is due to the deferral of uncertain epochs to a sleep expert. U-PASS presents a promising AI approach to incorporating uncertainty estimation in machine learning pipelines, improving their reliability and unlocking their potential in clinical settings.

CoRe-Sleep: A Multimodal Fusion Framework for Time Series Robust to Imperfect Modalities.

Sleep abnormalities can have severe health consequences. Automated sleep staging, i.e. labelling the sequence of sleep stages from the patient's physiological recordings, could simplify the diagnostic process. Previous work on automated sleep staging has achieved great results, mainly relying on the EEG signal. However, often multiple sources of information are available beyond EEG. This can be particularly beneficial when the EEG recordings are noisy or even missing completely. In this paper, we propose CoRe-Sleep, a Coordinated Representation multimodal fusion network that is particularly focused on improving the robustness of signal analysis on imperfect data. We demonstrate how appropriately handling multimodal information can be the key to achieving such robustness. CoRe-Sleep tolerates noisy or missing modalities segments, allowing training on incomplete data. Additionally, it shows state-of-the-art performance when testing on both multimodal and unimodal data using a single model on SHHS-1, the largest publicly available study that includes sleep stage labels. The results indicate that training the model on multimodal data does positively influence performance when tested on unimodal data. This work aims at bridging the gap between automated analysis tools and their clinical utility.

Sensory event-related potential morphology predicts age in premature infants.

OBJECTIVE: We investigated whether sensory-evoked cortical potentials could be used to estimate the age of an infant. Such a model could be used to identify infants who deviate from normal neurodevelopment. METHODS: Infants aged between 28- and 40-weeks post-menstrual age (PMA) (166 recording sessions in 96 infants) received trains of visual and tactile stimuli. Neurodynamic response functions for each stimulus were derived using principal component analysis and a machine learning model trained and validated to predict infant age. RESULTS: PMA could be predicted accurately from the magnitude of the evoked responses (training set mean absolute error and 95% confidence intervals: 1.41 [1.14; 1.74] weeks,p = 0.0001; test set mean absolute error: 1.55 [1.21; 1.95] weeks,p = 0.0002). Moreover, we show that their predicted age (their brain age) is correlated with a measure known to relate to maturity of the nervous system and is linked to long-term neurodevelopment. CONCLUSIONS: Sensory-evoked potentials are predictive of age in premature infants and brain age deviations are related to biologically and clinically meaningful individual differences in nervous system maturation. SIGNIFICANCE: This model could be used to detect abnormal development of infants' response to sensory stimuli in their environment and may be predictive of neurodevelopmental outcome.

Detecting Epileptic Seizures Using Hand-Crafted and Automatically Constructed EEG Features.

Epileptic seizure detection aims to replace unreliable seizure diaries by a model that automatically detects seizures based on electroencephalography (EEG) sensors. However, developing such a model is difficult and time consuming as it requires manually searching for relevant features from complex EEG data. Domain experts may have a partial understanding of the EEG characteristics that indicate seizures, but this knowledge is often not sufficient to exhaustively enumerate all relevant features. To address this challenge, we investigate how automated feature construction may complement hand-crafted features for epileptic seizure detection. By means of an empirical comparison on a real-world seizure detection dataset, we evaluate the ability of automated feature construction to come up with new relevant features. We show that combining hand-crafted and automated features results in more accurate models compared to using hand-crafted features alone. Our findings suggest that future studies on developing EEG-based seizure detection models may benefit from features constructed using a combination of hand-crafted and automated feature engineering.

Home recording of 3-Hz spike-wave discharges in adults with absence epilepsy using the wearable Sensor Dot.

OBJECTIVE: Home monitoring of 3-Hz spike-wave discharges (SWDs) in patients with refractory absence epilepsy could improve clinical care by replacing the inaccurate seizure diary with objective counts. We investigated the use and performance of the Sensor Dot (Byteflies) wearable in persons with absence epilepsy in their home environment. METHODS: Thirteen participants (median age = 22 years, 11 female) were enrolled at the university hospitals of Leuven and Freiburg. At home, participants had to attach the Sensor Dot and behind-the-ear electrodes to record two-channel electroencephalogram (EEG), accelerometry, and gyroscope data. Ground truth annotations were created during a visual review of the full Sensor Dot recording. Generalized SWDs were annotated if they were 3 Hz and at least 3 s on EEG. Potential 3-Hz SWDs were flagged by an automated seizure detection algorithm, (1) using only EEG and (2) with an additional postprocessing step using accelerometer and gyroscope to discard motion artifacts. Afterward, two readers (W.V.P. and L.S.) reviewed algorithm-labeled segments and annotated true positive detections. Sensitivity, precision, and F1 score were calculated. Patients had to keep a seizure diary and complete questionnaires about their experiences. RESULTS: Total recording time was 394 h 42 min. Overall, 234 SWDs were captured in 11 of 13 participants. Review of the unimodal algorithm-labeled recordings resulted in a mean sensitivity of .84, precision of .93, and F1 score of .89. Visual review of the multimodal algorithm-labeled segments resulted in a similar F1 score and shorter review time due to fewer false positive labels. Participants reported that the device was comfortable and that they would be willing to wear it on demand of their neurologist, for a maximum of 1 week or with intermediate breaks. SIGNIFICANCE: The Sensor Dot improved seizure documentation at home, relative to patient self-reporting. Additional benefits were the short review time and the patients' device acceptance due to user-friendliness and comfortability.

Microstate Analysis Reflects Maturation of the Preterm Brain.

Preterm neonates are at risk of long-term neurodevelopmental impairments due to disruption of natural brain development. Electroencephalography (EEG) analysis can provide insights into brain development of preterm neonates. This study aims to explore the use of microstate (MS) analysis to evaluate global brain dynamics changes during maturation in preterm neonates with normal neurodevelopmental outcome.The dataset included 135 EEGs obtained from 48 neonates at varying postmenstrual ages (26.4 to 47.7 weeks), divided into four age groups. For each recording we extracted a 5-minute epoch during quiet sleep (QS) and during non-quiet sleep (NQS), resulting in eight groups (4 age group x 2 sleep states). We compared MS maps and corresponding (map-specific) MS metrics across groups using group-level maps. Additionally, we investigated individual map metrics.Four group-level MS maps accounted for approximately 70% of the global variance and showed non-random syntax. MS topographies and transitions changed significantly when neonates reached 37 weeks. For both sleep states and all MS maps, MS duration decreased and occurrence increased with age. The same relationships were found using individual maps, showing strong correlations (Pearson coefficients up to 0.74) between individual map metrics and post-menstrual age. Moreover, the Hurst exponent of the individual MS sequence decreased with age.The observed changes in MS metrics with age might reflect the development of the preterm brain, which is characterized by formation of neural networks. Therefore, MS analysis is a promising tool for monitoring preterm neonatal brain maturation, while our study can serve as a valuable reference for investigating EEGs of neonates with abnormal neurodevelopmental outcomes.

The roles, challenges, and merits of the p value.

Since the 18th century, the p value has been an important part of hypothesis-based scientific investigation. As statistical and data science engines accelerate, questions emerge: to what extent are scientific discoveries based on p values reliable and reproducible? Should one adjust the significance level or find alternatives for the p value? Inspired by these questions and everlasting attempts to address them, here, we provide a systematic examination of the p value from its roles and merits to its misuses and misinterpretations. For the latter, we summarize modest recommendations to handle them. In parallel, we present the Bayesian alternatives for seeking evidence and discuss the pooling of p values from multiple studies and datasets. Overall, we argue that the p value and hypothesis testing form a useful probabilistic decision-making mechanism, facilitating causal inference, feature selection, and predictive modeling, but that the interpretation of the p value must be contextual, considering the scientific question, experimental design, and statistical principles.

Aggression Severity as a Predictor of Mortality in Dementia.

OBJECTIVES: In psychogeriatric units for patients with dementia and behavioral problems, aggression is prevalent. Predictions and timely interventions of aggression are essential to create a safe environment and prevent adverse outcomes. Our study aimed to determine whether aggression severity early during admission to these units could be used as an indicator of adverse outcomes. DESIGN: During one year, all aggressive incidents on a psychogeriatric unit were systematically recorded using the Revised Staff Observation of Aggression Scale (SOAS-R). The study investigated the link between the severity of incidents within the first 48 hours of admission and adverse outcomes. SETTING AND PARTICIPANTS: All patients included in the study were admitted to a psychogeriatric unit for dementia and behavioral problems between November 2020 and October 2021. METHODS: The study population was categorized into groups according to the level of aggression severity during the first 48 hours of admission. The impact of aggression severity on the duration of admission, aggression frequency and severity during admission, medication usage at discharge, discharge destination, and mortality risk were examined. RESULTS: During the initial 2 days of admission, 9 of 88 patients had 1 or more severe aggression incidents. An early manifestation of severe aggression was significantly associated with more incidents during hospitalization, a higher total SOAS-R score, and a sevenfold higher 1-year mortality risk compared with patients who did not or only mildly manifested aggression in the first 48 hours of admission. CONCLUSIONS AND IMPLICATIONS: An early manifestation of aggression not only poses a direct safety risk to all involved but is also an early indicator of patients at risk for more detrimental outcomes, specifically mortality risk. By identifying patients at higher risk for adverse outcomes early, health care providers can provide preventive or timelier interventions, mitigating the risk of adverse outcomes and optimizing care services.

Assessing a digital technology-supported community child health programme in India using the Social Return on Investment framework.

An estimated 5.0 million children aged under 5 years died in 2020, with 82% of these deaths occurring in sub-Saharan Africa and southern Asia. Over one-third of Mumbai's population has limited access to healthcare, and child health outcomes are particularly grave among the urban poor. We describe the implementation of a digital technology-based child health programme in Mumbai and evaluate its holistic impact. Using an artificial intelligence (AI)-powered mobile health platform, we developed a programme for community-based management of child health. Leveraging an existing workforce, community health workers (CHW), the programme was designed to strengthen triage and referral, improve access to healthcare in the community, and reduce dependence on hospitals. A Social Return on Investment (SROI) framework is used to evaluate holistic impact. The programme increased the proportion of illness episodes treated in the community from 4% to 76%, subsequently reducing hospitalisations and out-of-pocket expenditure on private healthcare providers. For the total investment of Indian Rupee (INR) 2,632,271, the social return was INR 34,435,827, delivering an SROI ratio of 13. The annual cost of the programme per child was INR 625. Upskilling an existing workforce such as CHWs, with the help of AI-driven decision- support tools, has the potential to extend capacity for critical health services into community settings. This study provides a blueprint for evaluating the holistic impact of health technologies using evidence-based tools like SROI. These findings have applicability across income settings, offering clear rationale for the promotion of technology-supported interventions that strengthen healthcare delivery.

Assessing focus through ear-EEG: a comparative study between conventional cap EEG and mobile in- and around-the-ear EEG systems.

Introduction: As our attention is becoming a commodity that an ever-increasing number of applications are competing for, investing in modern day tools and devices that can detect our mental states and protect them from outside interruptions holds great value. Mental fatigue and distractions are impacting our ability to focus and can cause workplace injuries. Electroencephalography (EEG) may reflect concentration, and if EEG equipment became wearable and inconspicuous, innovative brain-computer interfaces (BCI) could be developed to monitor mental load in daily life situations. The purpose of this study is to investigate the potential of EEG recorded inside and around the human ear to determine levels of attention and focus. Methods: In this study, mobile and wireless ear-EEG were concurrently recorded with conventional EEG (cap) systems to collect data during tasks related to focus: an N-back task to assess working memory and a mental arithmetic task to assess cognitive workload. The power spectral density (PSD) of the EEG signal was analyzed to isolate consistent differences between mental load conditions and classify epochs using step-wise linear discriminant analysis (swLDA). Results and discussion: Results revealed that spectral features differed statistically between levels of cognitive load for both tasks. Classification algorithms were tested on spectral features from twelve and two selected channels, for the cap and the ear-EEG. A two-channel ear-EEG model evaluated the performance of two dry in-ear electrodes specifically. Single-trial classification for both tasks revealed above chance-level accuracies for all subjects, with mean accuracies of: 96% (cap-EEG) and 95% (ear-EEG) for the twelve-channel models, 76% (cap-EEG) and 74% (in-ear-EEG) for the two-channel model for the N-back task; and 82% (cap-EEG) and 85% (ear-EEG) for the twelve-channel, 70% (cap-EEG) and 69% (in-ear-EEG) for the two-channel model for the arithmetic task. These results suggest that neural oscillations recorded with ear-EEG can be used to reliably differentiate between levels of cognitive workload and working memory, in particular when multi-channel recordings are available, and could, in the near future, be integrated into wearable devices.

Position paper From the digital twins in healthcare to the Virtual Human Twin: a moon-shot project for digital health research.

The idea of a systematic digital representation of the entire known human pathophysiology, which we could call the Virtual Human Twin, has been around for decades. To date, most research groups focused instead on developing highly specialised, highly focused patient-specific models able to predict specific quantities of clinical relevance. While it has facilitated harvesting the low-hanging fruits, this narrow focus is, in the long run, leaving some significant challenges that slow the adoption of digital twins in healthcare. This position paper lays the conceptual foundations for developing the Virtual Human Twin (VHT). The VHT is intended as a distributed and collaborative infrastructure, a collection of technologies and resources (data, models) that enable it, and a collection of Standard Operating Procedures (SOP) that regulate its use. The VHT infrastructure aims to facilitate academic researchers, public organisations, and the biomedical industry in developing and validating new digital twins in healthcare solutions with the possibility of integrating multiple resources if required by the specific context of use. Healthcare professionals and patients can also use the VHT infrastructure for clinical decision support or personalised health forecasting. As the European Commission launched the EDITH coordination and support action to develop a roadmap for the development of the Virtual Human Twin, this position paper is intended as a starting point for the consensus process and a call to arms for all stakeholders.

Machine learning to differentiate pulmonary hypertension due to left heart disease from pulmonary arterial hypertension.

Background and aims: Pulmonary hypertension due to left heart disease (PH-LHD) is the most frequent form of PH. As differential diagnosis with pulmonary arterial hypertension (PAH) has therapeutic implications, it is important to accurately and noninvasively differentiate PH-LHD from PAH before referral to PH centres. The aim was to develop and validate a machine learning (ML) model to improve prediction of PH-LHD in a population of PAH and PH-LHD patients. Methods: Noninvasive PH-LHD predictors from 172 PAH and 172 PH-LHD patients from the PH centre database at the University Hospitals of Leuven (Leuven, Belgium) were used to develop an ML model. The Jacobs score was used as performance benchmark. The dataset was split into a training and test set (70:30) and the best model was selected after 10-fold cross-validation on the training dataset (n=240). The final model was externally validated using 165 patients (91 PAH, 74 PH-LHD) from Erasme Hospital (Brussels, Belgium). Results: In the internal test dataset (n=104), a random forest-based model correctly diagnosed 70% of PH-LHD patients (sensitivity: n=35/50), with 100% positive predicted value, 78% negative predicted value and 100% specificity. The model outperformed the Jacobs score, which identified 18% (n=9/50) of the patients with PH-LHD without false positives. In external validation, the model had 64% sensitivity at 100% specificity, while the Jacobs score had a sensitivity of 3% for no false positives. Conclusions: ML significantly improves the sensitivity of PH-LHD prediction at 100% specificity. Such a model may substantially reduce the number of patients referred for invasive diagnostics without missing PAH diagnoses.

Acoustic stimulation as a promising technique to enhance slow-wave sleep in Alzheimer's disease: results of a pilot study.

STUDY OBJECTIVES: Sleep disturbances are common in people with Alzheimer's disease (AD), and a reduction in slow-wave activity is the most striking underlying change. Acoustic stimulation has emerged as a promising approach to enhance slow-wave activity in healthy adults and people with amnestic mild cognitive impairment. In this phase 1 study we investigated, for the first time, the feasibility of acoustic stimulation in AD and piloted the effect on slow-wave sleep (SWS). METHODS: Eleven adults with mild to moderate AD first wore the DREEM 2 headband for 2 nights to establish a baseline registration. Using machine learning, the DREEM 2 headband automatically scores sleep stages in real time. Subsequently, the participants wore the headband for 14 consecutive "stimulation nights" at home. During these nights, the device applied phase-locked acoustic stimulation of 40-dB pink noise delivered over 2 bone-conductance transducers targeted to the up-phase of the delta wave or SHAM, if it detected SWS in sufficiently high-quality data. RESULTS: Results of the DREEM 2 headband algorithm show a significant average increase in SWS (minutes) [t(3.17) = 33.57, P = .019] between the beginning and end of the intervention, almost twice as much time was spent in SWS. Consensus scoring of electroencephalography data confirmed this trend of more time spent in SWS [t(2.4) = 26.07, P = .053]. CONCLUSIONS: Our phase 1 study provided the first evidence that targeted acoustic stimuli is feasible and could increase SWS in AD significantly. Future studies should further test and optimize the effect of stimulation on SWS in AD in a large randomized controlled trial. CITATION: Van den Bulcke L, Peeters A-M, Heremans E, et al. Acoustic stimulation as a promising technique to enhance slow-wave sleep in Alzheimer's disease: results of a pilot study. J Clin Sleep Med. 2023;19(12):2107-2112.

L-SeqSleepNet: Whole-cycle Long Sequence Modeling for Automatic Sleep Staging.

Human sleep is cyclical with a period of approximately 90 minutes, implying long temporal dependency in the sleep data. Yet, exploring this long-term dependency when developing sleep staging models has remained untouched. In this work, we show that while encoding the logic of a whole sleep cycle is crucial to improve sleep staging performance, the sequential modelling approach in existing state-of-the-art deep learning models are inefficient for that purpose. We thus introduce a method for efficient long sequence modelling and propose a new deep learning model, L-SeqSleepNet, which takes into account whole-cycle sleep information for sleep staging. Evaluating L-SeqSleepNet on four distinct databases of various sizes, we demonstrate state-of-the-art performance obtained by the model over three different EEG setups, including scalp EEG in conventional Polysomnography (PSG), in-ear EEG, and around-the-ear EEG (cEEGrid), even with a single EEG channel input. Our analyses also show that L-SeqSleepNet is able to alleviate the predominance of N2 sleep (the major class in terms of classification) to bring down errors in other sleep stages. Moreover the network becomes much more robust, meaning that for all subjects where the baseline method had exceptionally poor performance, their performance are improved significantly. Finally, the computation time only grows at a sub-linear rate when the sequence length increases.

Predicting total lung capacity from spirometry: a machine learning approach.

Background and objective: Spirometry patterns can suggest that a patient has a restrictive ventilatory impairment; however, lung volume measurements such as total lung capacity (TLC) are required to confirm the diagnosis. The aim of the study was to train a supervised machine learning model that can accurately estimate TLC values from spirometry and subsequently identify which patients would most benefit from undergoing a complete pulmonary function test. Methods: We trained three tree-based machine learning models on 51,761 spirometry data points with corresponding TLC measurements. We then compared model performance using an independent test set consisting of 1,402 patients. The best-performing model was used to retrospectively identify restrictive ventilatory impairment in the same test set. The algorithm was compared against different spirometry patterns commonly used to predict restriction. Results: The prevalence of restrictive ventilatory impairment in the test set is 16.7% (234/1402). CatBoost was the best-performing machine learning model. It predicted TLC with a mean squared error (MSE) of 560.1 mL. The sensitivity, specificity, and F1-score of the optimal algorithm for predicting restrictive ventilatory impairment was 83, 92, and 75%, respectively. Conclusion: A machine learning model trained on spirometry data can estimate TLC to a high degree of accuracy. This approach could be used to develop future smart home-based spirometry solutions, which could aid decision making and self-monitoring in patients with restrictive lung diseases.

Personalized Longitudinal Assessment of Multiple Sclerosis Using Smartphones.

Personalized longitudinal disease assessment is central to quickly diagnosing, appropriately managing, and optimally adapting the therapeutic strategy of multiple sclerosis (MS). It is also important for identifying idiosyncratic subject-specific disease profiles. Here, we design a novel longitudinal model to map individual disease trajectories in an automated way using smartphone sensor data that may contain missing values. First, we collect digital measurements related to gait and balance, and upper extremity functions using sensor-based assessments administered on a smartphone. Next, we treat missing data via imputation. We then discover potential markers of MS by employing a generalized estimation equation. Subsequently, parameters learned from multiple training datasets are ensembled to form a simple, unified longitudinal predictive model to forecast MS over time in previously unseen people with MS. To mitigate potential underestimation for individuals with severe disease scores, the final model incorporates additional subject-specific fine-tuning using data from the first day. The results show that the proposed model is promising to achieve personalized longitudinal MS assessment; they also suggest that features related to gait and balance as well as upper extremity function, remotely collected from sensor-based assessments, may be useful digital markers for predicting MS over time.