A new 2026 study demonstrates that machine-learning models can use physiological indicators to predict relative cognitive performance groups in healthy adults.
In the September 2026 issue, Medicine & Science in Sports & Exercise published findings on whether physiological indicators can predict cognitive performance. The study is titled "From Fitness to Cognition: Machine-Learning Prediction of Cognitive Performance Using Physiological Parameters in Healthy Adults". It appears in volume 58, issue 9, spanning pages 2048 to 2062. The paper is available under the DOI 10.1249/MSS.0000000000004024. Researchers examined if cardiovascular data could feed interpretable machine-learning models to estimate relative cognitive performance.
The authors framed this work around the inherent difficulty of monitoring cognition frequently. Conventional cognitive tasks are highly burdensome for participants. They are not well suited to repeated assessment in high-pressure environments. Therefore, finding passive physiological markers offers a compelling alternative for monitoring mental readiness.
The analysis relied on a cross-sectional sample of 240 adults. Researchers recorded 39 physiological variables as model inputs to build a detailed cardiovascular profile. For the cognitive outcome, they used performance on the Trail Making Test. Completion time was divided at the median to form relatively shorter-time and longer-time groups.
The research design aimed to identify a compact set of physiological variables without relying on a single statistical technique. To achieve this, researchers combined four distinct feature-selection approaches. These methods included correlation, mutual information, genetic algorithms and recursive feature elimination. They paired these approaches with grid-tuned classifiers, stratified fivefold cross-validation and probability calibration.
This methodology helped researchers avoid building a model around arbitrary statistical noise. It allowed them to evaluate whether cardiac autonomic data could accurately distinguish between faster and slower cognitive test performers. The rigorous structure focused on establishing moderate discrimination capabilities rather than highly precise individual prediction. By using multiple statistical approaches, the authors improved the reliability of their underlying model structure.
Operators often struggle to determine when their teams are mentally prepared for demanding analytical work. The traditional approach relies heavily on subjective self-reporting or rigid schedules. This new study suggests a future where workload planning might be informed by objective physiological trends. If heart rate and autonomic data correlate with cognitive sharpness, leaders could potentially match critical tasks to peak biological readiness.
In practice, physical capacity directly influences our ability to sustain high-level mental output. During the toughest quarter of my career, I noticed that my ability to handle stress was directly tied to my cardiovascular fitness, not my mindset. I was trying to meditate my way out of a physiological deficit. Once we started looking at the data connecting aerobic capacity to emotional regulation and executive function, everything clicked. Physical capacity is the absolute foundation of mental resilience.
This connection highlights why building physical stamina is a professional requirement for sustainable executive performance. Incorporating exercise into a daily structure can provide measurable cognitive benefits over time. One meta-analysis of acute exercise reported a small positive overall effect on cognitive performance. This research showed an overall effect size of g = 0.097 across 1,034 effect sizes.
A separate 2026 meta-analysis found that high-intensity interval training improved executive function. That study reported a standardized mean difference of 0.38 following the specific exercise intervention. While this evidence focuses on active exercise interventions rather than passive monitoring, it supports the fundamental link between physiological health and cognitive capacity. Leaders should view regular movement as a structural tool to maintain energy and productivity throughout the workday.
The most successful approach in the study was a random forest model. This specific model used 10 features selected through recursive feature elimination. It outperformed an untuned logistic-regression baseline to deliver notable predictive results. The random forest model achieved 70.83% accuracy and an F1 score of 71.38%. It also reached an area under the receiver-operating-characteristic curve of 71.2%.
Researchers used SHAP-based interpretation to understand the internal logic of the model predictions. This method offers directional explanations rather than producing a completely opaque score. According to the SHAP analysis, older age, higher systemic vascular resistance and higher resting heart rate shifted predictions toward the longer-time group. These specific factors were associated with slower Trail Making Test performance.
Conversely, several unique physiological variables shifted predictions toward the shorter-time group. These included greater stroke volume, higher cardiac output, high-frequency heart-rate-variability power and respiratory sinus arrhythmia. This detailed breakdown provides a nuanced view of how cardiovascular health markers align with relative cognitive speed. The authors noted that several of these influential variables are potentially modifiable through focused lifestyle adjustments.
While the 70.83% accuracy represents a useful signal within the sample, it leaves substantial room for false positives. The model performed significantly better than chance-level classification. However, it is not accurate enough to function as a dependable cognitive readiness meter for individuals. A single prediction should not determine whether an executive is prepared to lead a critical meeting.
The current evidence presents several strict boundaries for practical corporate application. First, the analysis was entirely cross-sectional. The study identified predictive associations but did not track whether individual physiological changes preceded shifts in cognition. Therefore, the research does not prove that changing resting heart rate or cardiac output will directly improve cognitive performance.
The scope of the cognitive measurement was also highly specific to one testing format. The study predicted relative group membership based solely on Trail Making Test completion time. It did not measure broad intelligence, strategic judgment, creativity or complex decision-making under pressure. We cannot assume these results translate seamlessly to real-world productivity tasks like investment analysis or leadership negotiations.
Furthermore, the research relied on 39 physiological variables gathered in a controlled clinical setting. The authors specifically described the evaluation of wearable-accessible physiological features as a future validation step. They did not demonstrate that a commercial smartwatch can currently predict a worker's cognitive sharpness. A 2026 review confirmed that wrist-worn activity trackers are useful for many health-monitoring applications, particularly heart rate and step count. However, basic heart rate accuracy does not automatically validate complex derived physiological metrics.
Measurement reliability across diverse populations introduces another critical operational limitation. A 2026 scoping review found that the accuracy of photoplethysmography-based measurements can vary significantly between users. Factors including sex, age, body-mass index and skin tone may influence measurement quality. A model trained on one specific demographic might not generalize equally across an entire corporate workforce.
The study represents an important feasibility step between laboratory physiology and passive digital monitoring. The authors concluded that cardiovascular and autonomic variables showed a moderate ability to distinguish relative cognitive performance groups. The next meaningful research phase will involve testing whether wearable-accessible features can predict within-person cognitive changes. Researchers will need to evaluate these physiological trends across days and weeks under highly realistic working conditions.
This research aligns with a broader industry movement toward digital phenotyping. A 2026 scoping review specifically identified machine-learning efforts that use wearable-derived physiological features to predict cognitive scores. Some of this emerging work even includes people with mild cognitive impairment. As consumer hardware improves, the accuracy of specialized variables and derived readiness scores must be assessed independently from basic step tracking.
For executive teams, the defensible near-term use case involves voluntary self-experimentation and aggregated workload planning. A sensible performance approach integrates physiological trends with active sleep optimization and recovery data. It should also include subjective readiness scores, workload volume and actual work outcomes.
Organizations must prioritize strict privacy and data governance as these monitoring technologies mature over time. The present paper supports the feasibility of physiology-based cognitive assessment, but it does not justify employee surveillance. Automated judgments about individual capability, compensation or hiring decisions are absolutely not supported by the current science. We anticipate future enterprise systems will focus heavily on personalized calibration rather than universal readiness thresholds.
Stay connected for research and practical guidance on executive performance, energy, focus, sleep, recovery and longevity. Ideas built for people who want to stay sharp, capable and effective for the long run.
Build habits and systems that support clear thinking, steady energy and long term capacity throughout a demanding career.
explore the Blog
