A Dynamic Bayesian Network model for the simulation of Amyotrophic Lateral Sclerosis progression

Area of Science:

• Neuroscience
• Computational Biology
• Biostatistics

Background:

• Amyotrophic lateral sclerosis (ALS) is a fatal adult-onset neurodegenerative disease with heterogeneous clinical manifestations, making individual prognosis challenging.
• Current understanding of ALS progression and patient stratification is limited, hindering effective clinical practice and drug development.
• Predicting ALS progression and identifying patient subgroups are critical for advancing research and therapeutic strategies.

Purpose of the Study:

• To develop a Dynamic Bayesian Network (DBN) model for predicting amyotrophic lateral sclerosis (ALS) progression and identifying risk factors.
• To simulate the temporal evolution of ALS and predict patient survival and time to loss of vital functions.
• To stratify ALS patients into subgroups based on risk factors and disease progression for personalized medicine.

Main Methods:

• A Dynamic Bayesian Network (DBN) model was developed using data from over 4500 ALS patients from the Pooled Resource Open-Access ALS Clinical Trials Database (PRO-ACT).
• The DBN model was used to identify probabilistic relationships among clinical variables, risk factors, survival, and loss of vital functions.
• Simulations were performed to predict cohort survival, time to impairment of communication, swallowing, gait, and respiration, and to stratify patients by prognosis.

Main Results:

• The DBN model successfully predicted ALS trajectories, including survival and loss of autonomy in functional domains.
• Key biomarkers influencing survival time (e.g., bicarbonate, calcium) and interdependencies (e.g., phosphorus, movement, creatinine) were identified.
• The model enabled patient stratification into subgroups with different prognoses, analyzing the impact of specific variables on outcomes.

Conclusions:

• The developed DBN model offers a valuable tool for improving ALS prognosis and understanding disease heterogeneity.
• Risk factor analysis and simulation capabilities support a personalized medicine approach for ALS patient care.
• This approach provides deeper insights into ALS manifestations and facilitates targeted therapeutic strategies.