Real World Data Tool
I collaborated and presented project to stakeholders and customers. Designs are currently being socialized amongst senior leadership to influence upcoming roadmap.
Tempus at it’s core is a genetic testing company (think 23 & me but more clinical) This puts Tempus in a position where it holds a ton of patient’s de-identified molecular data and clinical data such as their genetic makeup, demographics, diagnosis and more importantly Patient Reported Outcomes (PRO)
Use Tempus's in house Molecular and Clinical Data to help clinicians make a data-driven and successful treatment plan for their patients with Neurological and Psychiatric conditions
Increase monthly test order rate from 875 to 1000
Increase PGx test order rate by introducing RWD which will be a key differentiator in a concentrated PGx market.
Increase adoption rate of Tempus Hub to 50%
Increase the adoption or usage of Tempus HUB, which houses RWD along with a number of Tempus offerings.
We knew we wanted to build a tool using clinical and molecular dataset to help clinicians make a better treatment plan. But what are our customers pain points?
To understand this, we first had to identify shortcomings in achieving a successful treatment plan. We conducted 5 user interviews with different User personas(Generalists, Experts, Nurse Practitioners, PCPs). We also used a platform called Sermo to gather quantitative research insights.
We started to ideate RWD with a strong hypothesis that drugs are selected through a process of elimination, putting the patients through excruciating emotional distress.
>70% of our clinicians said that their first treatment plans were either ineffective or harmful for their patients.
A clinician usually has 45 minutes to thoroughly understand and diagnose the patient's symptoms, review their medical history and create a treatment plan. Making it a herculean task to provide proper care for their patients.
Patients may become frustrated if they have unrealistic expectations about how quickly the medication will work or the extent to which it will alleviate their symptoms. Causing patients to stop taking medications prematurely.
Misdiagnosis can result in the prescription of medications or therapies that are not effective for the actual condition, potentially leading to unnecessary side effects or complications.
Use Tempus's in house Molecular and Clinical Data to help clinicians make a data-driven and successful treatment plan for their patients with Neurological and Psychiatric conditions
How might we increase clinician confidence and efficiency while creating a treatment plan?
How might we help clinicians manage patients' treatment expectation?
How might we expand the clinicians point of view with regard to the patient's diagnosis?
Fit within existing workflow
Clinicians time is precious therefore making the tool seamless so it fits within their workflow is key for it's success.
Accelerate finding answers
Reduce the number of clicks to get to the information by providing 'to-the-point' information
Data integrity
Ensure that the data is accurately and honestly represented by using the right data visualization
I facilitated a workshop with key stakeholders to identify key solutions we wanted to build in our MVP. The final doc was socialized amongst the the stakeholders after
The diagram below was shared with the company to share a high level functionality of the Real World Data tool. This shows what data points the algorithm will utilize to provide the results. And finally how this helps with push forward Tempus's mission
Optimal Treatment Options
Present a ranked list of most commonly used medications.
Problems it addresses
Trial and Error
Time on Treatment
Present how a drug affects this particular cohort. Show side effects, drug efficacy and molecular gene-drug interaction.
Problems it addresses
Unmanaged treatment expectations
Risk Prediction
Present a list of other co-morbidities that are closely related to the current diagnosis. Our hope is that this would act as a second opinion tool.
Problems it addresses
Misdiagnosis
Approach #1:
RWD exists as its own tab in the left navbar. This leads to a search screen where a user can input their specifications and get results.
We did not pursue this approach because:
A. This would be another workflow for our clinicians, taking valuable time away from their actual work. (Design Decision #1)
B. Since it is essentially a Search tool, it should be a subscription based payment model which would be a hard sell.
Approach #2:
RWD or what we are calling "Insights" exists as a tab within the Patient's profile. In this method the system auto populates the specifications for that particular patient and shares the results without additional input.
We pursued this approach because:
A. Calling it "Insights" humanizes the feature.
B. Gave us the opportunity to expand beyond RWD by providing insights into other offerings like Reports, PRO etc thereby increasing adoption rate (Business goal #2)
C. "Insights" is only visible for patients within Tempus HUB. As patients are only populated in the HUB when an order is placed, this would improve our test order rate. (Business goal #1)
Time on treatment is the Hero?
A user can land on the patient page and immediately look at the list of drugs on the left and it's cause and affect on the right. This approach has lesser clicks and more data in the first page.
On conducting usability tests it became clear that this was secondary information and doesnt not need to be front and center creating an information overload.
.png)
Medication list is the Hero?
Medication list is presented first. User can click into this list which then opens an overlay that consists of Time on Treatment.
This reduced information overload and created more space to add insights other than RWD.
.png)
Choosing a suitable React library helped scope the design for me. We chose Vizx for this particular feature, this greatly defined the style and visual components that I can use.
Finally, selecting the right data visualization chart that aligns with the needs of the our users and business alike.
Below is an example of different designs we ran through to identify the right data style to share the Data completeness chart. Data completeness is how comprehensive or whole the data we are summarizing is.
As a designer and researcher on this project that lacked a current user problem, I had to first validate our assumptions. In this case, the question we tried to answer was: Do our clinicians use Trial and Error to tailor a treatment plan for their patients? If yes, is this efficient?
Once we validated these questions, we were able to reframe our thinking and create a product with a unique value proposition.