Amyotrophic Lateral Sclerosis (ALS) is a progressive neurodegenerative disorder of adult onset, defined by the degeneration of both upper and lower motor neurons, ultimately resulting in functional decline and death. Disease progression involves a sequence of impairments across four distinct domains - walking/self-care, breathing, swallowing, and communication - culminating in complete functional loss and death. Due to its heterogeneity and non-linear clinical trajectory, with periods of stable disease preceded or followed by rapid decline in different functional domains, ALS represents a challenging use case for process mining methodologies, which enable the extraction and modeling of temporal patterns from clinical event logs. In this work, we present a hybrid framework that integrates process mining techniques with supervised machine learning models to predict patient-specific progression paths. The proposed approach employs First Order Markov Models (FOMM) for process discovery, coupled with multiclass classifiers, including decision trees, random forests, and XGBoost, to forecast the next functional impairment based on initial clinical profiles and previous clinical milestones. The study was conducted using the Pooled Resource Open-Access ALS Clinical Trials (PRO-ACT) dataset, the largest publicly ALS clinical trial repository available to date, demonstrating the feasibility and predictive utility of the model in characterizing ALS progression trajectories.
Predicting the Next Clinical Event in Amyotrophic Lateral Sclerosis using Process-Oriented Machine Learning Models: A Case Study
Erica Tavazzi;
2025
Abstract
Amyotrophic Lateral Sclerosis (ALS) is a progressive neurodegenerative disorder of adult onset, defined by the degeneration of both upper and lower motor neurons, ultimately resulting in functional decline and death. Disease progression involves a sequence of impairments across four distinct domains - walking/self-care, breathing, swallowing, and communication - culminating in complete functional loss and death. Due to its heterogeneity and non-linear clinical trajectory, with periods of stable disease preceded or followed by rapid decline in different functional domains, ALS represents a challenging use case for process mining methodologies, which enable the extraction and modeling of temporal patterns from clinical event logs. In this work, we present a hybrid framework that integrates process mining techniques with supervised machine learning models to predict patient-specific progression paths. The proposed approach employs First Order Markov Models (FOMM) for process discovery, coupled with multiclass classifiers, including decision trees, random forests, and XGBoost, to forecast the next functional impairment based on initial clinical profiles and previous clinical milestones. The study was conducted using the Pooled Resource Open-Access ALS Clinical Trials (PRO-ACT) dataset, the largest publicly ALS clinical trial repository available to date, demonstrating the feasibility and predictive utility of the model in characterizing ALS progression trajectories.
Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
