Coleman AI use cases

Predictive Recommendations For Safety Incidents

In this basic scenario, the AI generates recommendations for user responses to needlestick injuries at a clinic over time. Based on the actions of users, the AI learns the most frequently used next steps and makes recommendations to users in the future. The list of recommendations can evolve, depending on user actions. Specifically, they are subject to change, depending on the health components that are assigned. This example is specific to health care with needlestick injuries. But the concepts apply to other industries, based on their specific safety incidents.

Over a one-month period, 25 needlestick injuries are logged to Occupational Health. Users assigned one of three next steps to the employees who were impacted:

• In 15 of these, users assigned a bloodwork component to the injured party.
• In seven, users assigned a physical exam component.
• In three, users assigned an immediate urgent-care component.

As the components were assigned, the AI used the data to begin to build and prioritize recommendations. The list of prioritized components is presented to the user in a panel on the safety incident pages in Occupational Health. The user can review the list before determining the next steps and assigning a component the employee who were subject to needlesticks. In this scenario, the AI prioritizes the most frequently prescribed components to generate its list of recommendations. Based on this data, the bloodwork component has the highest priority for this time period. The physical exam component has the second-highest priority. The urgent-care component has the third-highest priority.

Near the end of the one-month period, users of Occupational Health at the clinic receive training for treatment of needlestick injuries. In this training, users learn that needlestick victims should go directly to urgent care for treatment. At that point, users begin to assign only the urgent-care health component to needlestick victims. Over the next month, 21 needlestick injuries were logged. All 21 were assigned the urgent-care health component.

As new data is added, the list of recommendations that is generated by the AI evolves.

• 24 users were assigned an immediate urgent-care component.
• 15 users were assigned a bloodwork component to the injured party.
• Seven users were assigned a physical exam component.

Now the list of recommended health components for needlestick injuries reads as follows:

• urgent care
• bloodwork
• physical exam

The list of recommendations continues to evolve indefinitely, based on the health components that users assign.