Aerial is a flexible workflow platform that displays all of the relevant objective and subjective treatment response data medical oncologists need to make more informed, more timely clinical decisions. A patient companion application collects information about treatment side effects and funnels this data into Aerial, providing medical oncologists with a more unified picture of patient response.
Inspired by a documentation model commonly used to communicate about patient progress, Aerial surfaces all of the clinical information medical oncologists need to make personalized treatment decisions.
Deciding what treatment to prescribe is far less challenging for medical oncologists than deciding when and why to continue, adjust, or change medications. Aerial helps care teams keep track of all three.
By pulling in and visualizing data from a patient companion application, Aerial gives care teams a holistic view of patient progress or deterioration. This allows care teams to more proactively manage side effects.
Aerial provides a central hub for care team communication, enabling medical oncologists, physician assistants, and nurses to collaborate on cases and discuss patient issues in context.
The treatment timeline visualizes past and current treatments, including drug dosages and frequencies. Medical oncologists can view treatments from pre-defined time ranges or select a custom range. All original scan, blood work, and procedure results are accessible from the timeline, and all dashboard components dynamically update as the medical oncologist scrubs across the timeline.
The 3D patient model provides a schematic representation of tumor or lesion size and location. The model changes to reflect remitting or progressing disease based on information gleaned from pathology reports. By clicking on the tumors or lesions, medical oncologists can view original reports. Vitals and detailed lab results reside below the patient model.
The cancer profile graph pulls and visualizes relevant lab results from the EMR. Graphs automatically update based on cancer and treatment type, so medical oncologists can monitor indicators specific to individual patients. Medical oncologists can easily observe treatment response trends in the context of ongoing regimens and treatment modifications.
The subjective patient response graph pulls in real-time data from the patient companion application. All symptoms are graphed on a 0 to 4 scale (i.e., none to disabling), which provides the data with structure and consistency. Checkboxes allow medical oncologists to flexibly choose which symptoms to visualize. Patient comments and photos submitted through the app can be accessed from the graph.
Aerial’s annotation feature ensures that rationale is always understood in context. Medical oncologists can isolate and comment on individual elements in the dashboard. This increases the likelihood that they will enter outcomes at the point of care rather than hours or days later. Comments can be toggled on or off, letting medical oncologists view or hide details on demand.
Medical oncologists emphasize treatment changes as important points of reference in their understanding of patient progress. Aerial clearly delineates these changes with a red dotted line. This draws medical oncologists’ attention to correlations between treatment decisions and patient improvement or deterioration.
Patients can input side effect information using slider controls. Aerial’s companion app asks patients about symptoms specific to their treatment. When a patient’s medication regimen changes, side effect options automatically update to reflect those most common for the new drug. Scores feed directly into the subjective patient response graph in the dashboard.
Patients can indicate location-specific symptoms, such as pain and neuropathy, on a simple graphic. Ideally, symptoms reported via this application feature would map directly to the 3D patient model in Aerial. Consistent capture of patient complaints allows medical oncologists to prioritize discussion points ahead of patient appointments.
Patients can attach photos of unanticipated symptoms and send comments directly to the care team, which are accessible from the subjective patient response graph in Aerial. When patients report a symptom rating of 3 or higher (i.e., severe to disabling), the application prompts them to notify their care team.
A collapsible side bar displays an overview of all discussion threads. Care team profile pictures accompany message notifications, so medical oncologists can always clearly identify the sender.
Care team structure can vary widely across practices and patients. Patient-specific care team information is stored in Aerial, so medical oncologists can call or send direct messages to care team members from the dashboard.
Nurse navigators have a unique landing page where they can view incoming patient comments and photos, take immediate action on issues, and communicate how they handled patient problems to the rest of the care team.
Our mission was to design a tool for medical oncologists that would: 1) facilitate information sharing; 2) support decision-making; and 3) promote empathetic communication between providers and patients. In service of this goal, we reviewed literature and conducted field research to better understand medical oncologists' ecosystem, workflow, and challenges. Through our research, we identified major touchpoints that medical oncologists have with other healthcare professionals and patients. We aimed to support a tailored approach by determining when and how aggregated diagnostic and treatment information should be consumed and communicated at each touchpoint.
We explored the people, technology, and decision-making processes involved in cancer care delivery. We asked medical oncologists and care team members to walk us through their daily workflow, which allowed us to find breakdowns and opportunities for improvement. We synthesized our research and built an affinity diagram, a bottom-up method that allowed us to find naturally occurring patterns in our data. Ultimately, we distilled our findings into 3 key areas of tension, which we used as a starting point for visioning and scoping.
We considered Roche's business interests, existing products, and the feasibility of our ideas given time constraints. We broadly chose to focus on data overload and lack of visualization given the overwhelming number of medical oncologists who reported these issues as huge hurdles in their workflow. Through our visioning session with Roche, we prioritized three areas for further exploration: 1) treatment visualization; 2) workflow streamlining; and 3) interface customization.
We knew that medical oncologists struggle to maintain control of their schedules, adapt existing tools to their needs, and aggregate relevant data from multiple sources. What we didn't yet know was how we could cohesively address these challenges. We returned to our earlier research to re-examine our most salient findings. With the below learning and guiding questions in mind, we considered how we could create maximum impact by addressing the greatest number of workflow painpoints.
1. Medical oncologists struggle to obtain a holistic view.
How might we help medical oncologists more easily access the data they need when they need it?
2. Medical oncologists can't find the needle in the haystack.
How might we make treatment response trends and outliers more readily apparent?
3. Patient self-reports are often incomplete or inaccurate.
How might we fill in the gaps when it comes to care team understanding of patient treatment experiences?
4. Treatment is a delicate balance between efficacy, preferences, and tolerance.
How might we help medical oncologists better understand and capture treatment tradeoffs and outcomes?
After four months of research, we were eager to brainstorm ideas for improving how medical oncologists experience data. We used the below techniques to narrow down to the core features that would give our minimum viable product (MVP)maximum impact.
We used several visioning techniques in preparation for storyboard creation. To stimulate creativity, we employed a strategy where we built upon "worst" solutions to current problems. We challenged ourselves to find workable, implementable solution components from these "terrible" ideas.
After our spring presentation, we worked with Roche to scope our opportunities down to 3 broad areas. We decided to pursue treatment visualization after creating a value matrix to better understand the overlap between Roche’s business objectives and user needs. We developed 7 storyboards to illustrate challenges and possible solutions related to treatment decision-making.
We tested our concepts with 6 medical oncologists to validate underlying needs and gauge reactions to potential solutions. We synthesized our findings and totaled the unique positive and negative feedback each concept received. We began developing low- and mid-fidelity prototypes that could serve as physical representations of the most popular concepts.
We prototyped in four phases using the following techniques: 1) paper prototyping; 2) low-fidelity wireframing; 3) mid-fidelity mockups (parallel prototyping); and 4) high-fidelity minimum viable product (MVP) development.
We tested our assumptions about information hierarchy and decision-making rationale. What level of detail should be provided up front? How is data visualization currently used by medical oncologists? How do medical oncologists decide between two seemingly equal treatment options?
We created 5 black and white screens in Sketch and linked them with InVision. We asked medical oncologists to give us feedback on the feasibility and usefulness of each prototype.