A new study demonstrates how AI-augmented care can improve clinical decision-making. Researchers tested an AI co-clinician system in a London hospital, focusing on sepsis detection in emergency patients.
The system analyzed patient data in real time, comparing results with doctors’ assessments. In 85% of cases, the AI matched or exceeded the accuracy of human clinicians. The study included 10,000 patient records from Imperial College Healthcare NHS Trust.
Lead researcher Dr. Alan Karthikesalingam emphasized the tool’s role as a decision-support system, not a replacement. The AI flagged potential sepsis cases within minutes, reducing the average detection time from over two hours to under 20 minutes. This early intervention could significantly lower mortality rates, according to the team.
The project was a collaboration between DeepMind Health and Imperial College London. Clinical trials began in 2021 and expanded to other UK hospitals in 2023. Regulatory approval is pending, with the team aiming for NHS integration by 2025.
Experts warn that AI must complement, not replace, doctor expertise. Privacy concerns around patient data also remain a key discussion point in the project’s next phase.
Source: deepmind.google