A current workflow for a clinic was tedious and not very convenient for their staff. They created a research proposal and invited the company to take part. I was the leader from our side for this project on and helped with the research question: Will a frail assessment completed via chatbot save the clinic time and money? 
How their problem was solved
The script 
Understanding what was required from the assessment was the first place I looked at. I wanted to make sure we are getting all the information needed in the assessment while sounding more conversational rather than just a form being filled out. 

The script went through many versions,  on a spectrum of conversational vs. Question/Answer:  
The different versions were created due to feedback by different parties.
The initial draft was sent to the doctors running this study. They wanted the questions to be more direct. 
The second draft was passed through those doctors as well as the head nurses at the clinic. 
The nurses thought it was too form-feeling and when there nurses are the ones that speak to the patient to fill this assessment in person, it is more of a conversation between the patient rather than straight Q&A. 
The third draft was sent to the doctors, the nurses, as well as a focus group that included a mixture of different nurses that run the assessment on a daily basis. This draft had almost the perfect balance, but some feedback we gathered was to make it more conversational. 
The final draft was then completed and used to assess patients. 
A working prototype on our staging environment was used to test the chat with different individuals. We tested it with our internal team, friends and family, and most importantly seniors. 
These tests were done observationally where I watched them to the assessment and take notes on what was simple and where an individual got stuck. I timed them doing the test to see an average time taken to complete the chat and reviewed the complete report that we create when a chat is complete. 
Small edits were made, such as the speed of the messages being sent, a couple content adjustments and a bug or two adjusted with dev help! 

The launch
Once this got into the hands of the real patients, my job was not over yet! I worked with the clinic to send the link to the patient, pass over all reports of the chat and associated it to the patient in their EMR. I recorded how long it took the users to complete the chats and followed up to make sure the patient didn’t run into any issues. 
About a quarter into the trial, we noticed one issue. This issue came up in conversation before the launch but was moved past because the researchers did not think it would come up. The issue? A lot of seniors don’t have/know how to properly use a smart phone. Because of this, we quickly maneuvered it to be both SMS messages and Email messages. Once we added this new medium, most of the users had no issues! 
The final
With the use of a chatbot to complete the assessment, we found that we did in fact save clinicians time. The time for patients did not change too much in regards to completing the assessment in person vs. in the chat, however, our goal was to save clinicians time and we achieved the goal. 
Unfortunately, there was not room in the clinic’s budget to continue using the chat to do their assessments, but they plan on possibly using it in the soon future. 

You may also like

Back to Top