To deliver effective healthcare a clinician needs to have accurate information about a patient so that the right treatment decisions can be made. The distributed, unstructured nature of medical records makes this a difficult task.
At the HLTH 23 healthcare innovation conference, Google announced new capabilities in its Vertex AI Search product that will solve this problem.
Vertex AI Search is an AI-powered search engine that enterprises can use to access their own structured and unstructured data.
Patient records often include test results, scans, doctor’s notes, and other FHIR data which is difficult for a doctor to interrogate. Vertex AI Search tuned for healthcare and life sciences is like a chatbot that knows everything about the patient.
A doctor could ask questions like, ‘Does the patient smoke?’ or ‘What medication has he taken over the last 12 months?’ and then instantly get the answers without having to wade through patient records.
The immediate benefit is that this AI solution can free up administrative tasks that keep many clinicians from actually providing treatment to their patients. A 2016 study funded by the American Medical Association found that doctors were spending 2 hours doing admin for every hour they spent with a patient.
Imagine the improvement in the level of healthcare a doctor could provide if they didn’t have to sift through patient records or find the right billing codes for a procedure.
Even better with Med-Palm 2
Vertex AI Search is great at understanding the intent behind a query and then extracting the right answer from a pile of data. Combining this with the power of Med-Palm 2, Google’s medically-tuned LLM, can help doctors make informed decisions faster and more efficiently.
The ultimate AI healthcare solution comes from leveraging a suite of tools Google has created.
- Google Cloud Healthcare Data Engine – Combines clinical data from different sources into one longitudinal patient record.
- BigQuery – The records are uploaded into BigQuery, Google’s data warehouse. BigQuery makes it easy to analyze or query the data.
- Vertex AI Search – Uses Google Healthcare API and AI to understand the intent of the query and extract the answer from the BigQuery ML model.
- Med-Palm 2 – Can analyze the patient information Vertex AI Search extracts and then deliver insights, and suggest diagnoses and treatments.
Healthcare providers like Mayo Clinic are already integrating the tool into some of their operations.
Vish Anantraman, CTO for Mayo Clinic, said “Healthcare is a data intensive industry with majority of the medical knowledge locked up in unstructured documents and text…New search functionality from Vertex AI provides Mayo Clinic new capabilities to use data to support a wide range of applications.”
Mayo Clinic is using the tool in administrative use cases, but not in clinical care yet. Healthcare providers are being suitably cautious in how much and how quickly patient care is entrusted to AI rather than a human doctor.
If these AI healthcare tools prove to be effective and reliable then we should see an accelerated broader rollout. AI could soon make it a lot easier for your doctor to provide you with better healthcare.
In the longer term, we could also see AI-powered self-care solutions that could save you a trip to the doctor altogether.
