Understanding palliative care trajectories through clustering, calibrated prediction models, and explainable…

AI / informaticsComputer methods and programs in biomedicine· Jul 2026

Understanding palliative care trajectories through clustering, calibrated prediction models, and explainable AI.

Battushig Migiddorj, Marijka Batterham, David Currow, Khin Than Win

Read paper

Palliative & End-of-LifeAI / informaticsQ1IF 6.7

Machine learning models predicted palliative care episode duration with only modest accuracy.

Understanding palliative care trajectories through clustering, calibrated prediction models, and explainable AI.

Battushig Migiddorj … Khin Than Win

Computer methods and programs in biomedicine · Jul 2026

Background

Despite growing demand for palliative care, access remains limited and often occurs late in life.

Purpose

To (1) identify distinct patient subgroups based on symptom burden and functional status, and (2) predict palliative care episode duration using the identified subgroups.

Methods

A retrospective cohort study was conducted using national data from the Australian Palliative Care Outcomes Collaboration, including adults (≥18 years) who died between 2014 and 2023 (261,290 patients).

n = 261,290 patients

R² ≈ 0.2

Results

R² ≈ 0.2

duration was hard to predict from these factors alone

n = 261,290 patients

More results

Calibration improved agreement between predicted and observed outcomes, particularly for longer episodes.

In contrast, clustering did not improve predictive accuracy but enhanced interpretability by summarizing patient heterogeneity.

More results

Prediction accuracy varied systematically across clusters, diagnostic groups, and episode duration, with higher accuracy in shorter, late-stage episodes and greater uncertainty in longer trajectories.

Functional status and episode type were the dominant drivers of care duration.

“

Conclusion · 1 of 3

Palliative care needs follow a continuous severity gradient, with prediction accuracy varying across disease stages, most closely aligned to functional status.

Conclusion · 2 of 3

Prediction is most accurate near the end of life and least accurate earlier, highlighting a clinical paradox where intervention potential is greatest when predictions are most uncertain.

Conclusion · 3 of 3

Methodologically, these findings show that calibration improves prediction reliability, while clustering enhances interpretability without increasing accuracy.

Understanding palliative care trajectories through clustering, calibrated prediction models, and explainable AI.

Understanding palliative care trajectories through clustering, calibrated prediction models, and explainable AI.

Related papers