The Medical Futurist discusses several astonishing discoveries made by artificial intelligence (AI) in the field of medicine, highlighting AI’s ability to detect unusual associations and provide insights that are beyond human capabilities.
Key points from the article include:
Detecting Patients’ Race from Medical Images: AI algorithms developed by MIT scientists can accurately predict the self-reported race of patients from medical images alone, a task that even seasoned physicians cannot do. Despite extensive research, the exact mechanism by which the AI achieves this remains a mystery.
Diagnosing Type 2 Diabetes from Voice Analysis: A study published in Mayo Clinic Proceedings: Digital Health demonstrated that AI models could analyze short voice recordings to diagnose type 2 diabetes with better-than-chance accuracy. This suggests the potential for future applications where devices like smartphones could detect health conditions through voice patterns.
Brain Waves for Antidepressant Treatment: AI has been used to analyze brainwaves to identify the best antidepressant for patients. A study showed that 65% of patients with a specific brainwave pattern responded well to the antidepressant sertraline, indicating that AI can significantly improve treatment selection.
Identifying Rare Diseases from Photos: AI-based software has been trained to identify rare hereditary diseases from portrait photos of patients. This method has shown high accuracy rates in identifying the disease-causing gene, which could expedite diagnosis and treatment for rare conditions.
Predicting Recovery from Coma: AI developed by the Chinese Academy of Sciences and PLA General Hospital in Beijing can predict with high accuracy whether patients in a coma or vegetative state will regain consciousness, even contradicting doctors’ predictions in some cases.
Detecting Alzheimer’s Before Symptoms Manifest: Researchers trained an AI algorithm to detect Alzheimer’s disease from FDG-PET brain scans years before symptoms appear. The AI showed 100% sensitivity in detecting the condition on average more than six years prior to diagnosis.
Assessing Heart Attack Risk from Retinal Images: Google researchers used AI to analyze retinal images and identify signs indicating long-term cardiovascular risks. This method could potentially enable quick screening tests for assessing cardiovascular risk.
AI for Hospital Admission Predictions: A pilot project in England uses AI to predict which patients might need hospital admission. The AI allocates points based on health conditions and contributing factors, aiding GPs in early intervention and resource planning for hospitals.
The article concludes that while these AI discoveries are promising, further validation and large-scale studies are needed. Nonetheless, AI’s role in detecting risks, aiding in diagnosis, and taking preventive measures is becoming increasingly integral in healthcare.