UCSF Study Tests AI Chatbot to Help Women Recognize Heart Disease

UC San Francisco researchers are testing HeartBot, a conversational AI app designed to help women recognize heart disease risks and heart attack warning signs. Led by nurse scientist Yoshimi Fukuoka, a professor in the UCSF School of Nursing, the project was developed by an interdisciplinary team from UCSF, UCLA and UC Davis. It first launched as a text message service in 2021 and has evolved into HeartBot II, a 2025-upgraded chatbot powered by a large language model and available on iOS and Android for trial participants. The app answers questions about heart disease, symptoms, risk factors and prevention in real time, using a curated knowledge bank rather than open-ended AI output to reduce the risk of false or misleading medical advice. Conversations are personalized and can recall prior details, such as a family member’s heart attack. Researchers shaped the language around how women speak about health, replacing medical terms such as risk factors with everyday wording after users reported confusion. The project addresses a public health gap. Heart disease is the leading cause of death among women in the United States, affecting more than 60 million women, with the highest prevalence among non-Hispanic Black women. According to the American Heart Association, one in three women dies from heart disease, yet awareness has fallen sharply. The share of women who knew it was the top killer dropped from 65 percent in 2009 to 44 percent in 2019, especially among Black, Hispanic and younger women. HeartBot emphasizes that women’s heart attack symptoms can be less typical than chest pain and may include shortness of breath, nausea, vomiting, back pain or jaw pain. The goal is to encourage users to seek emergency care, especially by calling 911, when symptoms appear. UCSF is now conducting a six-month randomized controlled trial with 201 female UCSF Health patients. Half receive immediate access to HeartBot II educational sessions, while the control group gains access later. Researchers will measure whether conversations improve long-term retention of heart disease information and symptom recognition. If efficacy and safety are established, the team plans to offer HeartBot to the general public and expand access through QR codes in clinics and public spaces.