Ready2Care

Inspiration

Training incoming caretakers is time-consuming. Rather than being able to take more clients, nurses at healthcare agencies have to ensure that new caretakers perform all of the required skills properly and safely. We built EmpowerCare so that AI can do that work for them. In the end, the nurse will still be the one to make the final decision on whether a caretaker can successfully perform the skill, but they will only see the attempt after AI has cleared them first, saving them a lot of time that they would otherwise spend correcting the caretakers.

What it does

EmpowerCare makes training caretakers efficient by allowing nurses to assign safety maneuvers for caretakers to practice, such as safely assisting someone from a bed to a wheelchair. The caretaker watches the provided demonstration, then records themself performing the assigned maneuver and receives a pass-fail grade. If they fail, they can watch the demonstration again and keep retrying the maneuver until they do it successfully. Once they pass, the recording of the successful attempt is sent to the nurse to verify. The nurse can override the AI's grade if needed.

How we built it

We used Gemini API to analyze the videos. We used Python and Typescript to code this tool. We used Tiger Data to store our data.

Challenges we ran into

One challenge we have come across so far is optimizing API usage. Video ingestion via Gemini uses a lot more API credits than we thought, so we had to be careful with how/when we utilized this during development and testing.

Accomplishments that we're proud of

So far, we have a working product with pages for both nurses and caregivers in training. It can successfully analyze videos of caretakers attempting the assigned skills, and provides accurate feedback with the exact timestamps of where the feedback applies.

What we learned

We have learned a lot about how healthcare agencies operate as well as AI video analyzation.

What's next for EmpowerCare

With enough employees using our program, something we would want to add in the future is a way for clients to be automatically matched with specific caregivers based on how well their needs match the caregiver's skills.