The temporal event-based model: Learning event timelines in progressive diseases.

Area of Science:

• Biostatistics
• Computational Biology
• Neuroscience

Background:

• Progressive diseases are characterized by event timelines (e.g., symptom onset, biomarker changes).
• Accurate prediction of event timing is crucial for early clinical trial interventions.
• Existing models lack the ability to estimate time intervals between disease events, providing only event order.

Purpose of the Study:

• To introduce a novel probabilistic model, the Temporal Event-Based Model (TEBM), for inferring disease progression timelines.
• To demonstrate the TEBM's capability in analyzing sparse and irregularly sampled biomarker data.
• To validate the TEBM's performance in neurodegenerative diseases like Alzheimer's disease (AD) and Huntington's disease (HD).

Main Methods:

• Developed a probabilistic model (TEBM) to infer timelines of biomarker events.
• Applied the TEBM to analyze sparse, irregularly sampled datasets from AD and HD.
• Validated model performance using external datasets and compared it with existing models.

Main Results:

• The TEBM accurately recapitulates known event orderings in AD and HD.
• The model provides novel estimations for timescales between consecutive biomarker events.
• TEBM demonstrated improved performance over current models, enabling better patient stratification and enhancing clinical trial power.

Conclusions:

• The Temporal Event-Based Model (TEBM) is a powerful tool for inferring disease progression timelines from biomarker data.
• TEBM offers significant advantages in understanding disease dynamics and improving the efficiency of clinical trials for progressive conditions.
• The model's applicability extends beyond neurodegenerative diseases to various progressive conditions.