Parallel teamed up with Decision Labs to design and build a web-based simulation tool that allows pharmaceutical sales teams to better demonstrate the benefits of specific Sodium-glucose Cotransporter-2 (SGLT2) inhibitors to the National Health System decision-makers. The primary target was to demonstrate the reduction of hospital admissions that is achieved by increasing the use of a new inhibitor.
The web-based application is centred around a simulation model that visualises how patients with a mix of type 2 diabetes, heart failure, and chronic kidney disease move through the health system, known as the patient pathway. One of the main benefits is that users can optimise the model to visualise the beneficial impact, such as increasing the adoption of health checks, increasing the adoption of a specific SGLT inhibitor, and reducing outpatient appointments by 25% through patient-initiated follow-ups.
The challenge presented was to improve the user experience of the base simulation output from the third-party software, AnyLogic platform. The initial simulation and its settings were presented in the AnyLogic interface. The user experience was found to be unworkable.
An early example of the simulated model
Salespeople needed to allow for highly detailed input parameters and fine-tune the simulation model to demonstrate the benefit of specific inhibitors. This workflow became highly complex, with up to 100 input parameters and a lot of variability across different patient populations for each locality. Therefore, salespeople and NHS officials needed an intuitive interface to collaboratively fine-tune both input and outcome parameters to avoid errors and ensure the accuracy of the simulation.
To solve this challenge, we created a clear flow that allowed salespeople to define up to 100 input and outcome parameters. These parameters were compiled into logical groupings, making it easy for users to work through them. To avoid false simulations and wasted time, we also created an alert system alongside the groupings that would flag any issues, either by missing inputs or erroneous inputs.