DDIRC
Drug-drug interaction risk calculator
Project description
When someone is ill, often prescribe a drug.
To cure the illness.
Consider that the world population is getting older elderly people with multiple 'issues' that need treatment. Nowadays, numerous drugs are being administered to patients because of multiple diseases. In the USA, 9% of people over 55 take ten or more prescription drugs, and they can interact with each other, causing an increase or decrease in the efficacy or toxicology of the drug. In Europe, 197,000 deaths per year are attributed to drug-drug interactions. To protect patient safety and improve Pharma R&D efficiency, pharmaceutical developers must anticipate potential drug-drug interactions early during development.
The drug-drug interaction risk calculator (DDIRc) is a tool integrated into PharmaPendium. PharmaPendium (PP) helps customers benchmark their drug candidates against approved and marketed drugs to quickly understand if something is a blocker for approval. It is a risk assessment tool for translational safety, drug efficacy, and DMPK (Drug metabolism and pharmacokinetics).
This page is a short story of how we designed/improved the researchers' workflow.
![Timeline](https://ahmetbektes.com/wp-content/uploads/2024/01/Timeline.jpg)
This tool was developed in collaboration with Sanofi, Merck, Servier, Boerhinger Ingelheim and others with the aim of improving patient safety. In return for their time, they will have early access and a say in the development. In the early stages, we created a user journey (including pain and gain) together with possible actions. We later updated this deliverable to define the scope of the first releases. Below, you can see the basic end-to-end flow.
![DDI-basicflow](https://ahmetbektes.com/wp-content/uploads/2024/01/DDI-basicflow.png)
Project page:
Compared to the old design, the simulations are stored under projects, and experts can run more than one simulation and compare the results.
![sidemenu-topmenu](https://ahmetbektes.com/wp-content/uploads/2024/01/ddi-projects.jpg)
Input form:
This is one of the most complicated input forms I have encountered; it contains so much logical routing. We worked together on the input form with Tina Bota, a designer teammate. We aimed;
- Improving the order (wizard with indentation)
- Guiding the user (gradual progression and science)
- Declutter the information detours (some data is required to calculate another data point. If the user doesn't have it, the system will inject some default value to predict interactions)
- Be transparent (communicate the sources of the value and how it is being calculated)
![DDI-form](https://ahmetbektes.com/wp-content/uploads/2024/01/DDI-form-.png)
![sidemenu-topmenu](https://ahmetbektes.com/wp-content/uploads/2024/01/ddi-form3.jpg)
Prediction results and visualizations:
Result pages contain different visualizations for different purposes. The experts can see overall interactions on the drug class page (landing page). If they want to go deep into drugs and see exact values, they can go to a forest plot or table. After two user research studies, most improvement points were related to this area.
![sidemenu-topmenu](https://ahmetbektes.com/wp-content/uploads/2024/01/ddi-jelly3.jpg)
Acting as UX Researcher?
While designing the product, our team didn't have UX research resources. Due to my previous experience and education, I led, set up, and conducted two UX studies. I want to share this experience with you in the following session.
Alpha test set-up:
We started with an observing session – we didn't guide them during the first 30 minutes. In the first half, we gave the tool and did not talk; we observed the users and how they interacted with the system. Depending on their interest, we conducted the moderated session in the second half. Victim form and perpetrator form procedures were different. We used quota sampling; we tried to have equal tests for the form.
Beta test set-up:
In Beta testing, we needed to revisit some essential decisions, mainly Navigation, Filtering, and Forest Plot Visualization. Different alternatives were created and tested.
Results were shared in the format below – and these findings were prioritized and later included in the development pipeline.
![Wall](https://ahmetbektes.com/wp-content/uploads/2024/01/Wall.png)
![results.png-2](https://ahmetbektes.com/wp-content/uploads/2024/01/results.png-2.png)
![Excel-scorecard2](https://ahmetbektes.com/wp-content/uploads/2024/01/Excel-scorecard2.jpg)
![prioritization](https://ahmetbektes.com/wp-content/uploads/2024/01/prioritization.png)
![Jira](https://ahmetbektes.com/wp-content/uploads/2024/01/Jira.png)
Reflection
DDIRc was a strategically crucial piece for PharmaPendium. PharmaPendium was built upon content and data extraction. The second layer is creating valuable predictions for the customers rather than just providing information. This logic extended into other predictive tools, which helped several companies during their drug approval procedures.
In general, I don't think conducting UX studies on your design is healthy. However, I was working on increasing design maturity. I created a standard in the business unit and shared my templates with the other designers in UX Showcase sessions. Another reflection point is the fluentness of conducting remote user studies. Logistics of the user studies can take time and effort. Although we didn't have a budget for a UX researcher, I was able to utilize the Health Team's UXR tools.
Date
2020