Beyond Volume: A Pathophysiology-Driven Approach to Evaluating Physical Treatments for Oedema of Lymphatic Origin — Reflections of the Belgian Society of Lymphology
DOI: 10.29245/2767-5122/2026/3.1174 View / Download PdfBourgeois Pierre*1, Adriaenssens Nele*2,3, Barbieux Romain4, Roman Mirela M5, Harfouche Joseph6, Leduc Olivier4
1Services of Dermatology and Vascular Surgery, Erasme and HIS-IZZ Hospitals, Université Libre de Bruxelles, Brussels, Belgium.
2Rehabilitation Research, Vrije Universiteit Brussel, Brussels, Belgium.
3Medical Oncology Department, Universitair Ziekenhuis Brussel, Brussels, Belgium.
4Lympho-Phlebology Unit, Department of Occupational and Environmental Physiology, Haute Ecole HE2B ISEK, Brussels, Belgium.
5Department of Mammo-Pelvic Surgery, CHU St Pierre, Université Libre de Bruxelles, Brussels, Belgium.
6Department of Physical Therapy, Delta Hospitals, Brussels, Belgium
Know Thy Limits: Calibrated Uncertainty for Safer Rehabilitation
Joel Strickland
Intellegens, The Studio, Chesterton Mill, Cambridge, UK
Rehabilitation clinicians make threshold-based decisions—exercise progression, return-to-work clearance, heat safety management—that depend on reliable physiological monitoring. Yet current noninvasive methods for estimating core body temperature provide only point predictions with no indication of reliability. A prediction of 38.3°C tells clinicians nothing about whether the true value is 38.2°C or 39.0°C, making it impossible to distinguish a trustworthy estimate from a dangerously uncertain one. This limitation is especially concerning in rehabilitation, where autonomic dysfunction, cardiovascular medications, and atypical physiology cause prediction errors to vary unpredictably across patients and conditions. This mini-review argues that calibrated uncertainty—not accuracy alone—should be a foundational requirement for physiological monitoring in rehabilitation. It introduces conformal prediction, a framework that produces statistically valid prediction intervals: when configured for 95% confidence, the true temperature falls within the stated bounds approximately 95% of the time. Proof-of-concept evidence spanning over 140,000 measurements across six high-heat operational domains demonstrates that such calibration is technically achievable in real-world settings. For rehabilitation practice, uncertainty-aware monitoring enables risk-stratified exercise progression, defensible return-to-work decisions grounded in explicit confidence bounds, and scalable telemonitoring with transparent escalation pathways. The central principle is that uncertainty should be treated as a first-class clinical signal—used to pace progression, trigger conservative action, and distinguish high-risk physiology from low-trust measurement.
DOI: 10.29245/2767-5122/2026/3.1173 View / Download Pdf