What can a box of colliding particles tell us about a patient in intensive care?
The connection is not clinical but conceptual. Work recognized by Yu Deng's 2026 Fields Medal helped establish how the microscopic dynamics of many colliding particles can give rise to a macroscopic statistical law, the Boltzmann equation.
That achievement has renewed attention to one of science's most enduring questions: How do countless interactions at small scales produce order, disorder and collective behavior at larger scales?
An Editorial published online on July 14, 2026, in Intelligent Medicine recently brings a related question into critical care. Its authors propose that critical illness may arise not simply when organs fail, but when the body loses its ability to regulate physiological disorder across immune, circulatory, metabolic and cellular systems.
The Entropic Critical Illness Theory, or ECIT, views living organisms as open, non-equilibrium systems. A healthy organism continuously generates and dissipates entropy while maintaining structured, adaptive regulation. From this perspective, the problem in critical illness is not entropy itself, but the failure to keep entropy production physiologically organized.
From organ dysfunction to loss of coordination
ECIT does not replace diagnoses such as sepsis, acute respiratory distress syndrome, acute kidney injury or shock. Instead, it asks whether they may also reflect a deeper, convergent loss of system-wide coordination.
According to the theory, infection, trauma or another acute insult can disrupt the body's multilevel regulatory architecture, while host factors and medical interventions may alter the trajectory.
The resulting disorder is described through two linked latent constructs.
Host response entropy (HRE) refers to loss of informational structure and coordination across inflammatory, immune, coagulation, metabolic and neuroendocrine responses. It does not simply mean that inflammation is high: a strong response may still be organized and adaptive.
Hemodynamic entropy (HDE) describes loss of order in blood flow and oxygen delivery. Global blood pressure may appear acceptable while microcirculatory flow remains uneven and tissue-level oxygen delivery is impaired.
ECIT proposes that HRE and HDE reinforce one another. Dysregulated host responses can damage the endothelium and disturb microvascular control; impaired perfusion and hypoxia can then amplify cellular stress and inflammatory dysregulation.
The consequences converge on the "critical unit," a terminal microcirculatory–mitochondrial functional unit in which oxygen delivery and cellular energy production must remain closely coupled. Dysfunction at this level may contribute to progressive multi-organ dysfunction.
Beyond normalizing isolated numbers
Correcting one abnormal bedside value does not necessarily restore the underlying physiology. Raising blood pressure may not normalize microcirculatory perfusion, while lowering an inflammatory marker may not re-establish coordination across host-response networks.
ECIT therefore reframes existing critical care practice. It organizes treatment around three linked aims: control the primary insult; restore coordination within the host response; and improve blood flow and oxygen distribution so delivery better matches cellular demand.
Existing interventions would be judged not only by short-term changes in isolated variables, but also by whether they move the patient toward a more coordinated physiological state.
From outcome prediction to physiological state estimation
This systems view may offer a clinically interpretable research direction for artificial intelligence. HRE and HDE cannot currently be measured as single bedside thermodynamic quantities.
Future models would instead need to infer these hidden states from longitudinal relationships among physiological waveforms, variability measures, inflammatory and metabolic markers, lactate and perfusion trajectories, vasopressor requirements, organ-support intensity and microcirculatory data where available.
The goal would not be to produce one universal "entropy score," or merely predict death or organ dysfunction from a snapshot. An entropy-informed AI system might estimate whether physiological organization is improving or deteriorating, identify transitions from adaptive to maladaptive responses, and assess whether an intervention is restoring coordination or contributing to further disruption.
Such models would need interpretable outputs, robust longitudinal datasets, prospective testing and external validation across intensive care units and patient populations.
A theory to be tested
ECIT will ultimately be judged by whether it generates testable hypotheses. Future studies must determine whether HRE and HDE can be measured reliably, add value beyond established clinical indicators and inform decisions that improve care.
More broadly, the framework challenges researchers to look beyond isolated abnormalities and examine how physiological relationships deteriorate over time. Its central question is both scientific and clinical: how early can medicine recognize that a living system is losing coherence, and what would it take to help restore it? In this sense, the authors propose a conceptual evolution from "critical care medicine" to "critical illness medicine," broadening the field beyond the management of established organ dysfunction to understanding and modifying the dynamic systemic processes that govern the onset, progression and potential reversibility of critical illness.
Reference
DOI: https://doi.org/10.1016/j.imed.2026.07.003
About the Journal
Intelligent Medicine is a peer-reviewed, open-access journal focusing on the integration of artificial intelligence, data science, and digital technology in clinical medicine and public health. The journal has a latest JCR Impact Factor of 7.8 and an Elsevier CiteScore of 16.5, reflecting its growing international influence. It is published by the Chinese Medical Association in partnership with Elsevier. To learn more about Intelligent Medicine, please visit: https://www.sciencedirect.com/journal/intelligent-medicine
