Synthetic longitudinal clinical data, with static and temporal mixed-type components, can help unlock large-scale deep learning models to tackle complex diseases. However, learning to generate realistic patients faces dual challenges: modeling the inherently complex structure of longitudinal data and protecting patient privacy. We introduce PatientFlow, a generative modeling method combining Variational Autoencoders for data representation with Flow Matching for patient generation. We extensively evaluated the generative model on a longitudinal cohort of patients with Amyotrophic Lateral Sclerosis (N = 1560) using both qualitative and quantitative methods. The ability of the method to generate realistic patient data, further validated by expert clinicians, shows its potential application to other diseases. Prognostic models trained on synthetic data across five clinically relevant endpoints matched and sometimes outperformed the models trained on real data. Our results demonstrate that PatientFlow can effectively model longitudinal clinical data with high fidelity, opening promising avenues for sharing and augmenting datasets for deep learning applications in healthcare without compromising privacy.

Progressive neurological conditions like Amyotrophic Lateral Sclerosis (ALS) require regular and detailed monitoring to track functional decline and support timely clinical decisions. Traditional in-clinic assessments, however, can be burdensome and infrequent, often missing daily symptom variability. Smartphones and wearables offer a promising alternative for remote data collection. We present HomeSenseALS, a smartphone-based digital phenotyping application developed for ALS patients and caregivers, using a user-centered design approach involving patients, caregivers, and clinicians. HomeSenseALS collects multimodal data, including self-reported data from the Revised Amyotrophic Lateral Sclerosis Functional Rating Scale (ALSFRS-R), speech and respiratory recordings, and human movement passive data. To assess its potential for tracking ALS progression, we conducted a cross-sectional in-clinic study with 27 ALS patients and a longitudinal at-home study with 11 ALS patients over 21 weeks. Self-reported ALSFRS-R scores showed a strong correlation with in-clinic assessments. High-level physical activity features, such as time spent indoors, were associated with gross motor function. Speech analysis differentiated patients with and without bulbar dysfunction and correlated with respiratory parameters. Home spirometry aligned with clinical respiratory measures. These findings support the feasibility of using HomeSenseALS for remote, multimodal monitoring of ALS progression, potentially reducing patient burden while enabling more continuous and granular tracking of functional decline.

Amyotrophic Lateral Sclerosis (ALS) is a fast-acting neurodegenerative disease, characterized by loss of muscle movement and heterogeneity in disease evolution.This poses a challenge in predicting the best time for therapy administration.Here, we propose Deep Temporal Consensus Clustering (DTCC), a stratification method to uncover patient groups with similar disease progression.Using only the initial 6-month follow-up period, DTCC uncovered five clusters that were evaluated in terms of disease evolution and time-to-event.For three critical events (non-invasive ventilation, gastrostomy and death) the attained groups show distinct 10year progressions, validating the approach.

BACKGROUND: Amyotrophic Lateral Sclerosis (ALS) is a fatal neurodegenerative disorder characterised by the progressive loss of motor neurons in the brain and spinal cord. The fact that ALS's disease course is highly heterogeneous, and its determinants not fully known, combined with ALS's relatively low prevalence, renders the successful application of artificial intelligence (AI) techniques particularly arduous. OBJECTIVE: This systematic review aims at identifying areas of agreement and unanswered questions regarding two notable applications of AI in ALS, namely the automatic, data-driven stratification of patients according to their phenotype, and the prediction of ALS progression. Differently from previous works, this review is focused on the methodological landscape of AI in ALS. METHODS: We conducted a systematic search of the Scopus and PubMed databases, looking for studies on data-driven stratification methods based on unsupervised techniques resulting in (A) automatic group discovery or (B) a transformation of the feature space allowing patient subgroups to be identified; and for studies on internally or externally validated methods for the prediction of ALS progression. We described the selected studies according to the following characteristics, when applicable: variables used, methodology, splitting criteria and number of groups, prediction outcomes, validation schemes, and metrics. RESULTS: Of the starting 1604 unique reports (2837 combined hits between Scopus and PubMed), 239 were selected for thorough screening, leading to the inclusion of 15 studies on patient stratification, 28 on prediction of ALS progression, and 6 on both stratification and prediction. In terms of variables used, most stratification and prediction studies included demographics and features derived from the ALSFRS or ALSFRS-R scores, which were also the main prediction targets. The most represented stratification methods were K-means, and hierarchical and expectation-maximisation clustering; while random forests, logistic regression, the Cox proportional hazard model, and various flavours of deep learning were the most widely used prediction methods. Predictive model validation was, albeit unexpectedly, quite rarely performed in absolute terms (leading to the exclusion of 78 eligible studies), with the overwhelming majority of included studies resorting to internal validation only. CONCLUSION: This systematic review highlighted a general agreement in terms of input variable selection for both stratification and prediction of ALS progression, and in terms of prediction targets. A striking lack of validated models emerged, as well as a general difficulty in reproducing many published studies, mainly due to the absence of the corresponding parameter lists. While deep learning seems promising for prediction applications, its superiority with respect to traditional methods has not been established; there is, instead, ample room for its application in the subfield of patient stratification. Finally, an open question remains on the role of new environmental and behavioural variables collected via novel, real-time sensors.

Longitudinal cohort studies to study disease progression generally combine temporal features produced under periodic assessments (clinical follow-up) with static features associated with single-time assessments, genetic, psychophysiological, and demographic profiles. Subspace clustering, including biclustering and triclustering stances, enables the discovery of local and discriminative patterns from such multidimensional cohort data. These patterns, highly interpretable, are relevant to identifying groups of patients with similar traits or progression patterns. Despite their potential, their use for improving predictive tasks in clinical domains remains unexplored. In this work, we propose to learn predictive models from static and temporal data using discriminative patterns, obtained via biclustering and triclustering, as features within a state-of-the-art classifier, thus enhancing model interpretation. triCluster is extended to find time-contiguous triclusters in temporal data (temporal patterns) and a biclustering algorithm to discover coherent patterns in static data. The transformed data space, composed of bicluster and tricluster features, capture local and cross-variable associations with discriminative power, yielding unique statistical properties of interest. As a case study, we applied our methodology to follow-up data from Portuguese patients with Amyotrophic Lateral Sclerosis (ALS) to predict the need for non-invasive ventilation (NIV) since the last appointment. The results showed that, in general, our methodology outperformed baseline results using the original features. Furthermore, the bicluster/tricluster-based patterns used by the classifier can be used by clinicians to understand the models by highlighting relevant prognostic patterns.

Amyotrophic Lateral Sclerosis is a devastating neurodegenerative disease causing rapid degeneration of motor neurons and usually leading to death by respiratory failure. Since there is no cure, treatment's goal is to improve symptoms and prolong survival. Non-invasive Ventilation (NIV) is an effective treatment, leading to extended life expectancy and improved quality of life. In this scenario, it is paramount to predict its need in order to allow preventive or timely administration. In this work, we propose to use itemset mining together with sequential pattern mining to unravel disease presentation patterns together with disease progression patterns by analysing, respectively, static data collected at diagnosis and longitudinal data from patient follow-up. The goal is to use these static and temporal patterns as features in prognostic models, enabling to take disease progression into account in predictions and promoting model interpretability. As case study, we predict the need for NIV within 90, 180 and 365 days (short, mid and long-term predictions). The learnt prognostic models are promising. Pattern evaluation through growth rate suggests bulbar function and phrenic nerve response amplitude, additionally to respiratory function, are significant features towards determining patient evolution. This confirms clinical knowledge regarding relevant biomarkers of disease progression towards respiratory insufficiency.

Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disease causing patients to quickly lose motor neurons. The disease is characterized by a fast functional impairment and ventilatory decline, leading most patients to die from respiratory failure. To estimate when patients should get ventilatory support, it is helpful to adequately profile the disease progression. For this purpose, we use dynamic Bayesian networks (DBNs), a machine learning model, that graphically represents the conditional dependencies among variables. However, the standard DBN framework only includes dynamic (time-dependent) variables, while most ALS datasets have dynamic and static (time-independent) observations. Therefore, we propose the sdtDBN framework, which learns optimal DBNs with static and dynamic variables. Besides learning DBNs from data, with polynomial-time complexity in the number of variables, the proposed framework enables the user to insert prior knowledge and to make inference in the learned DBNs. We use sdtDBNs to study the progression of 1214 patients from a Portuguese ALS dataset. First, we predict the values of every functional indicator in the patients' consultations, achieving results competitive with state-of-the-art studies. Then, we determine the influence of each variable in patients' decline before and after getting ventilatory support. This insightful information can lead clinicians to pay particular attention to specific variables when evaluating the patients, thus improving prognosis. The case study with ALS shows that sdtDBNs are a promising predictive and descriptive tool, which can also be applied to assess the progression of other diseases, given time-dependent and time-independent clinical observations.