AI-Assisted Personalized Heat-Risk Alerts

Extreme heat poses increased health risks for adults living with chronic conditions. Conventional heat-health warning systems generally provide population-level advisories and may not account for individual clinical vulnerability. This study will evaluate an artificial intelligence-assisted personalized heat-risk alert system designed to integrate individual clinical characteristics with environmental exposure information.

The trial will enroll 120 adults with hypertension, type 2 diabetes, chronic kidney disease, cardiovascular disease, and/or obesity from the outpatient department of a selected tertiary-care hospital in Gujranwala, Pakistan. Participants will be randomized 1:1 to an intervention or control group and followed for six weeks.

On days when the Pakistan Meteorological Department same-day forecast maximum temperature is ≥36°C, the intervention system will process prespecified clinical characteristics and the temperature forecast through a locked AI Prediction Model. Participants will be classified into low, moderate, or high heat-related acute clinical-event risk categories, with corresponding personalized heat-health messaging. The control group will receive a generic PMD heat-health advisory through the same patient-facing mobile application without AI-based risk stratification or clinical personalization.

The primary outcome is Heat-Related Illness Symptom Score (HRISS) at Week 6. Secondary outcomes include heat-protective behaviors, heat-health knowledge, attitudes and practices, heat-related emergency department visits and hospital admissions, and application engagement. The AI model will be developed and internally validated using a separate historical hospital dataset containing heat-related emergency department visits/admissions and will be locked before intervention delivery.

Inclusion criteria

• Adults aged 18 years or older attending the outpatient department of the selected tertiary-care hospital during the recruitment period.
• Have a documented diagnosis of at least one chronic non-communicable disease associated with increased susceptibility to heat-related illness, including hypertension, type 2 diabetes mellitus, chronic kidney disease, cardiovascular disease, or obesity (BMI ≥30 kg/m²).
• Have access to a personal smartphone capable of receiving study heat-risk alert notifications.
• Be able to read Urdu or English, or have a household member/caregiver available to read and explain study alerts when required.
• Be willing and able to provide written informed consent.
• Intend to remain within the study catchment area for the duration of the six-week study period to facilitate follow-up.

Exclusion criteria

• Patients requiring immediate emergency treatment or hospital admission at the time of recruitment.
• Individuals with severe cognitive impairment, dementia, psychotic illness, or another medical condition that limits their ability to understand study procedures or provide informed consent.
• Patients with terminal illness or those receiving palliative care.
• Individuals with severe visual, hearing, or communication impairments that prevent effective receipt of the study alert intervention and outcome assessment without a reliable caregiver.
• Pregnant women, because pregnancy has distinct physiological responses to heat exposure and would require separate clinical risk stratification beyond the scope of this study.
• Participants currently enrolled in another clinical trial or structured behavioral intervention related to heat-health, climate adaptation, or chronic disease self-management.
• Participants who are unable or unwilling to comply with study procedures or complete the required follow-up assessment.