Precision Exercise Programming in Optimizing Cardiometabolic Health and Physical Performance through Artificial Intelligence Assisted Training

Figure Ilustration AI

FORMOSA NEWS - Surabaya - AI-Driven Exercise Programs Outperform Conventional Workouts in Boosting Heart Health and Fitness. Artificial intelligence can now tailor physical exercise to an individual’s exact physiological responses, dramatically enhancing cardiorespiratory fitness and kardiometabolic health compared to traditional workout routines. According to a landmark 2026 study published by Dr. Irmantara Subagio, Sulton Nur Arifin, Egy Raafi Alafian, and Ary Nindya Kuswara from Universitas Negeri Surabaya in Indonesia, dynamic, AI-assisted training programs yield significantly superior physiological adaptations across all key health markers. The findings arrive at a critical moment when global physical inactivity is driving a surge in non-communicable diseases, proving that data-driven exercise prescription is the key to optimizing human health and physical performance.

The Rising Crisis of Inactivity and the Need for Precision Exercise

Physical inactivity has reached alarming levels worldwide. Global health data indicates that over 31% of the adult population fails to meet minimum physical activity guidelines, significantly elevating the risk of hypertension, obesity, insulin resistance, and cardiovascular diseases. While exercise is widely prescribed to combat these conditions, human bodies react to physical exertion in vastly different ways. Standardized, "one-size-fits-all" fitness programs often fail because they do not account for individual biological variations. A workload that causes optimal adaptation in one person might cause overtraining or yield no benefits in another. To bridge this gap, the researchers at Universitas Negeri Surabaya set out to evaluate whether integrating wearable sensors with artificial intelligence algorithms could dynamically adjust exercise intensity, volume, and recovery to deliver superior health outcomes.

How the Study Was Conducted
The research team conducted a 12-week parallel-group randomized controlled trial (RCT) involving 120 adult participants. The trial directly compared an AI-assisted precision exercise program against a traditional, non-personalized exercise routine. Participants were randomly allocated into two equal groups:

• AI-Assisted Group: Participants wore digital fitness sensors that tracked real-time physiological indicators. An AI system analyzed data such as heart rate, training load, and fatigue to dynamically adjust exercise volume and intensity for each session.
  
• Conventional Group: Participants followed a standard, fixed physical exercise program with periodic manual check-ins but without real-time algorithmic adjustments.
  

To measure progress, the researchers recorded baseline data and post-intervention metrics after 12 weeks. Evaluated parameters included maximum oxygen consumption ($\text{VO}_2\text{max}$), resting heart rate, systolic and diastolic blood pressure, fasting blood glucose, body composition, and physical strength and endurance tests. Data were statistically evaluated using a mixed-model analysis of variance (ANOVA).

Key Research Findings
After 12 weeks of training, the AI-assisted group demonstrated statistically significant and superior improvements across every measured cardiometabolic and physical performance indicator compared to the conventional group:

• Cardiorespiratory Fitness ($\text{VO}_2\text{max}$): The AI-assisted group achieved a $\text{VO}_2\text{max}$ increase of $+4.20\text{ ml/kg/min}$ (rising from $33.15$ to $37.35\text{ ml/kg/min}$), compared to an increase of $+2.47\text{ ml/kg/min}$ in the conventional group.
  
• Cardiovascular Health: Systolic blood pressure in the AI group dropped by $7.90\text{ mmHg}$ (versus $3.90\text{ mmHg}$ in the control group), while diastolic blood pressure decreased by $4.95\text{ mmHg}$ (versus $1.97\text{ mmHg}$). Resting heart rate also saw a major drop of $7.53\text{ bpm}$ in the AI group compared to $3.97\text{ bpm}$ in the conventional group.
  
• Metabolic Control: Fasting blood glucose levels in the AI group decreased by $-7.86\text{ mg/dL}$ (falling from $100.78$ to $92.93\text{ mg/dL}$), outperforming the $-4.71\text{ mg/dL}$ reduction seen in the conventional group.
  
• Body Composition: Participants in the AI program lost an average of $2.68\text{ kg}$ in body weight, reduced body fat by $2.08\%$, shed $3.55\text{ cm}$ in waist circumference, and gained $0.94\text{ kg}$ of lean muscle mass—all surpassing the results of the control group.
  
• Muscle Strength and Endurance: Handgrip strength in the AI group improved by $+4.15\text{ kg}$, push-up capacity increased by $+6.65\text{ repetitions}$, and physical endurance (time-to-exhaustion) expanded by $+2.16\text{ minutes}$.
  

Real-World Impact and Practical Applications
The implications of this research extend far beyond academic literature, offering practical solutions for digital health platforms, athletic coaches, public health officials, and clinical practitioners. By replacing static workout templates with dynamic AI adaptation, individuals can achieve maximum physiological gains in less time while reducing the risk of exercise-induced injuries or burnout. Furthermore, integrating AI-driven exercise algorithms into commercial smartwatches and fitness wearables could democratize access to personalized sports medicine, making preventive healthcare more accessible to the general public.

Author Profile
Dr. Irmantara Subagio, S.Pd., M.Pd. Universitas Negeri Surabaya (UNESA), Indonesia. Expertise: Sports Science, Physical Education, Precision Exercise Programming, and Fitness Technology
Co-Authors: Sulton Nur Arifin, Egy Raafi Alafian, and Ary Nindya Kuswara (Universitas Negeri Surabaya)

Source
Irmantara Subagio, Sulton Nur Arifin, Egy Raafi Alafian, dan Ary Nindya Kuswara. Precision Exercise Programming in Optimizing Cardiometabolic Health and Physical Performance through Artificial Intelligence Assisted Training. Formosa Journal of Applied Sciences (FJAS), Vol. 5, No. 8, Tahun 2026, Halaman 1797-1814
DOI : https://doi.org/10.55927/fjas.v5i8.98
URL : https://journalfjas.my.id/index.php/fjas