Manipal Hospitals uses Gemini to cut nurse handover time from 90 minutes to 20

The problem

Manipal Hospitals serves nearly 7 million patients a year across 37 hospitals. When a nurse's shift ends, they must pass the patient's condition and care plan to the incoming nurse. With more than 10,500 beds, the volume of data made this slow, and the hospital says nurses routinely added about 90 minutes to their shifts to write and read the reports. It links this to fatigue, mistakes and lower job satisfaction.

What they built

A tool that produces a handover summary on demand. When a nurse asks for one, the system pulls the patient's records and has Google's Gemini model write a short report in the ISBAR format (Identify, Situation, Background, Assessment, Recommendation), a standard structure for clinical handovers.

How they did it

Patient data from the hospital's TrakCare record system flows in near real time to a data store on Google Cloud. A time-based filter picks only the information relevant to the nurse's shift, and only that structured data goes to Gemini on Vertex AI. Manipal calls this controlled generation and says it improves the accuracy and consistency of the report; it gives no figures for errors or invented content. Google Cloud Consulting led the build from idea to production, starting with a minimum viable product and holding more than eight rounds of sessions with nurses. A pilot ran at one hospital before the rollout. A Google press release says Deloitte supported Manipal on the wider programme.

Results

Manipal reports that handover time dropped from 90 minutes to 20. The tool is live in 23 hospitals and used by more than 5,000 nurses every day. The hospital says the time saved lets nurses focus on direct patient care and that the process is less exposed to errors from handwritten notes and fatigue.