Blog

What Would You Do with 30,000 Hours of Extra Time?

Share:

ref-auto-icons.png

Imagine what could happen if staff had an extra 25,000 hours annually and clinicians had 5,000 hours of time per year – they could spend more time focused on patient care. The DPE product team is one step closer to making that a reality with Referrals Automation 2.0.

UC San Francisco receives about 7 million fax pages annually that require manual sorting, organizing, and data entry. These repetitive and time-consuming tasks could be better allocated to improving our patients’ overall experience and care.

Referrals Automation 2.0 will use artificial intelligence (AI) to extract relevant information from faxes to reduce – and eventually nearly eliminate – staff time to read and process referral faxes and other documents like laboratory and clinical notes.

In June, we began experiments and will evaluate results after four to six weeks. So far, results look promising and our next step will be to release a departmental pilot.

Once the product is launched, the AI learning model will essentially “practice till perfect.” That is, compared to Referrals Automation 1.0 which has a set performance and improves in a stepwise manner, the performance of 2.0 is continuously improving. How? As staff identify errors and omissions in the information extraction performed by the AI, corrections are made and then fed back into the model making it more accurate over time. For example, staff may have to initially review information in a data entry field 100 percent of the time. As the machine learns, that review time will progressively decrease and eventually be eliminated. As each data field is unique and varies in complexity (e.g. an address versus a diagnosis code), the time to reach the optimal accuracy level is also variable.

Referrals Automation 2.0 is slated to launch in FY22. Stay tuned and keep track of our progress in upcoming newsletters.

Want to learn more? Check out Just 3 Questions with product manager Ramki Yerramsetty and data scientist Lu Chen. Ramki and Lu provide their insights on the evolution of Referrals Automation and why it is needed, explain how AI systems learn and improve accuracy over time, and share their excitement for the way AI can revolutionize this aspect of healthcare operations.