Tashkent, Uzbekistan (UzDaily.uz) — Artificial intelligence can help patients understand their health conditions and prepare for medical appointments, but final medical decisions must remain with healthcare professionals, experts said at a discussion in Tashkent held alongside the presentation of the STADA Health Report 2026.
The discussion featured Gulnara Dzhambekova, deputy director for research at the International Centre for Molecular Allergology of the Academy of Sciences of Uzbekistan and a professor at Tashkent State Medical University; Nilufar Nosirjonova, director of strategic development for artificial intelligence and machine learning at Sino AI; Shavkat Musinov, Uzbekistan champion and healthy lifestyle practitioner; and digital healthcare expert Farhad Furkatov.
The discussion drew on findings from the STADA Health Report 2026. According to the study, 55% of Uzbekistan’s residents turn to artificial intelligence for health-related questions. Another 45% would consider replacing a doctor’s visit with an AI consultation under certain circumstances, although 75% still prefer an in-person consultation with a physician. Meanwhile, 92% of respondents said they self-medicate in at least one situation.
Why Patients Stop Treatment
Dzhambekova identified one of the main risks of self-medication as patients independently discontinuing treatment for chronic conditions, particularly bronchial asthma. When they begin to feel better, they stop taking their medication. By the time they return to their doctor, the previous treatment regimen may no longer be effective, requiring the treatment to be reviewed and intensified.
There are several reasons for this behaviour, she said. Doctors may not have enough time to explain treatment fully, while patients may lack the money to continue therapy or the means to travel to appointments. Some also distrust prescribed medication or believe their condition is more serious than the diagnosis suggests.
In such cases, Dzhambekova stressed, a face-to-face conversation is essential for reassuring patients, explaining the treatment and establishing a partnership. Artificial intelligence cannot replace this interaction, she said.
She also noted that patients increasingly arrive at consultations with interpretations of their symptoms and test results generated by AI.
Chatbots That Agree With Everything
Nosirjonova warned that chatbots, particularly free models, can be prone to agreeing with users. If someone asks whether they should take an antibiotic, a model may simply confirm a decision the person has already made. This, she said, is a model failure.
A good AI tool should follow three principles, Nosirjonova explained. First, it should not automatically agree with the user but should ask follow-up questions, such as how high the patient’s temperature is and how long the symptoms have lasted. Second, it should help people understand what may be happening without making decisions for them. Third, it should recognise “red flags” and advise users to seek urgent medical attention when dangerous symptoms appear. No model is 100% accurate, she added.
Nosirjonova identified two main areas where AI can be useful: understanding and decision-making. AI can patiently explain medical information, repeat explanations as often as necessary and help people prepare for a doctor’s appointment. However, “Decision-making is always the doctor’s job,” she said.
Medical chatbots differ from general-purpose models because they rely on verified, regularly updated sources and are tested with medical professionals, Nosirjonova said. General-purpose models can produce fabricated information.
Language is another challenge. Models are trained predominantly on English-language material, while Uzbek-language data remains limited, making responses in Uzbek less accurate. Nosirjonova advised users to provide as much detail as possible when asking questions. “A detailed question is already a detailed answer,” she said.
Where Is the Line Between Self-Medication and Medical Care?
According to the study, 49% of Uzbekistan’s residents attribute self-medication to their own experience and knowledge of what works for them. Meanwhile, 22% reported having had a negative experience with self-medication, more than twice the average across the 20 countries surveyed, at 9%.
Dzhambekova said self-care can be appropriate in certain circumstances. Staying home during a viral respiratory infection may be preferable to waiting in a doctor’s queue, as it reduces the risk of infecting others. People who know they have an allergy may take an anti-allergy medication themselves. However, she warned that anaphylaxis is a medical emergency in which every second counts and taking risks is unacceptable.
Commenting on the 22% figure, Nosirjonova said it demonstrated that AI should be used to improve understanding rather than make medical decisions.
Musinov advised people to pay attention to signals from their bodies. If mild discomfort occurs during exercise, they should reduce the intensity, for example by switching from running to walking slowly. If the symptoms persist or recur, they should consult a specialist.
Prevention Starts With One’s Surroundings
Obesity and unhealthy lifestyles were identified by 44% of respondents as one of the country’s main healthcare challenges, ranking second after environmental factors. The study found that 16% of respondents undergo all available preventive health screenings, while another 36% complete some of them.
Musinov said people often seek medical help only after a problem has become serious, even though prevention is easier. He said habits take 21 days to form.
People frequently use AI to create exercise plans, but it cannot replace a trainer, he said. After a workout, AI will not offer personal encouragement, which can be a strong source of motivation for many people.
Addressing nutrition, Musinov noted that local cuisine tends to be high in fat and that people are often insufficiently active, making weight gain difficult to avoid without regular exercise.
Social surroundings help people maintain healthy habits, he said. “Our environment changes us.” When a trainer or friends are waiting for someone at a morning workout, skipping it becomes more difficult.
Musinov began running in 2019 after taking part in an international half-marathon in Samarkand. Today, his friends and former classmates also run and set an example for their children. He views the growing number of participants in running events as a sign that interest in a healthy lifestyle is becoming more firmly established.
Dzhambekova agreed that healthy habits are formed primarily within families and called for greater health education, including at school.
Wearables and Health Data
The study found that 89% of Uzbekistan’s residents use at least one tool to monitor their health, while 25% use mobile health applications. Experts agreed that such data can be useful but must be interpreted with caution.
Dzhambekova warned against placing complete trust in readings from wearable devices. A gadget may show normal indicators, but eating certain foods before physical activity can trigger anaphylaxis, which the device cannot predict.
Musinov said his team’s trainer monitors how athletes sleep and how they feel in the morning. Before competitions, the athletes follow a routine of sleeping eight hours a night for three days.
Furkatov said digital services should help people quickly obtain initial information about their health and direct them to a doctor when serious symptoms arise.
Smartwatches and mobile applications that combine sleep and activity data can help people make more informed decisions, he said. His team also uses gamification to encourage physical activity, organising challenges and gatherings where participants exercise and socialise together.
The experts also discussed the security of medical data. Furkatov said it was too late to focus on fears about data collection alone; the priority should be learning how to use services properly and carefully checking what information is being shared and with whom.
Nosirjonova identified a lack of transparency as a major concern, noting that users often do not know who has access to their data.
Concluding the discussion, the participants agreed that AI can be a useful tool for understanding health issues, but it cannot replace a doctor. Final medical decisions must remain with qualified professionals.
