Dozee unveils landmark study on AI-powered early warning system

Dozee, a leader in health AI in India, has revealed the findings of a study published in the journal Frontiers in Medical Technology. This study, conducted at King George’s Medical University (KGMU), is one of the largest observational studies of its kind in Indian tertiary care. It highlights the impact of Dozee’s AI-powered Early Warning System (EWS), which can predict patient health deterioration up to 16 hours in advance. This capability provides healthcare professionals with an opportunity to intervene early and potentially save lives.

In India, approximately 1.9 million patients in general wards rely on manual spot checks for monitoring, despite the presence of 2 million hospital beds. Dozee’s AI-Powered Remote Patient Monitoring and EWS offers a solution that has the potential to transform care across 95 per cent of hospital capacity. This technology delivers continuous monitoring that ensures comprehensive healthcare at a lower cost compared to ICU services.

The observational study monitored over 700 patients across 85,000 hours. It demonstrated that Dozee’s Continuous Contactless Remote Patient Monitoring and EWS can improve traditional manual processes. By delivering alerts up to 16 hours before a critical event, Dozee’s system allows healthcare professionals to take action earlier, which can enhance patient outcomes. The study indicated that healthcare practitioners could save an average of 2.4 hours per day through this system. Key metrics such as alert sensitivity, specificity, and average time from initial alert to deterioration were analysed, providing evidence of Dozee’s effectiveness.

In many hospitals in India, continuous monitoring is typically limited to ICUs. This leaves general wards, where most patients are, at risk of undetected clinical deterioration. The study shows that Dozee’s EWS bridges this gap by tracking vital signs, including heart rate, respiratory rate, and blood pressure. Results revealed that Dozee’s EWS predicted patient deterioration in 67 per cent to 94 per cent of cases, enabling healthcare providers to intervene before conditions became critical. This early detection has the potential to save 2.1 million lives annually and reduce healthcare costs by ₹6400 crores.

Dr Himanshu Dandu, a Professor in the Department of Medicine at KGMU, highlighted the technology’s potential. “This system enables early detection and continuous patient monitoring, providing a scalable and affordable solution tailored to the demands of healthcare systems facing heavy patient loads. The ability to detect signs of patient health deterioration can significantly improve their survival rates,” he said.

Dr Jean-Louis Teboul, an intensivist and critical care expert from Paris-Saclay University, noted the global implications of the study. “What we have achieved in India has the potential to reshape healthcare globally. The challenges may differ, but the need for equitable, timely, and affordable care remains universal,” he stated.

Gaurav Parchani, CTO and Co-Founder of Dozee, expressed confidence in the findings. “The results of this study affirm what we’ve always believed—this real-world evidence demonstrates technology’s ability to transform healthcare, making it more efficient, accessible, and equitable. We’re not just solving a problem for India but laying the groundwork for global healthcare solutions,” he said.

The study was authored by a team of experts, including Dr Himanshu Dandu and Dr Ambuj Yadav from KGMU, alongside Dozee’s clinical research team, which includes Gaurav Parchani, Dr Kumar Chokalingam, and Ms. Pooja Kadambi. It also featured contributions from Dr Rajesh Mishra, a former ISCCM President, and Dr Ahsina Jahan from Bangladesh. The study has gained international recognition, including input from Dr Jean-Louis Teboul and Dr Jos M. Latour from the University of Plymouth in the UK. The findings indicate that Dozee’s health AI addresses global healthcare gaps and offers a solution that could serve as a model for global adoption.