Cancer Informatics Clinical Decision Support
Clinical Decision Support
A tool for oncologists at the point of care.

The Problem
How do you help a general Oncologists navigate an EMR to obtain the data they need, along with the latest science, and seamlessly integrate with their workflow at the point of care?

In oncology, there tends to be two groups of physicians, specialists and generalists. Specialists are highly trained, up-to-date on the latest science, and familiar with all the options—specifically for their narrow specialty. They need to compete with other specialists in their area of expertise and are often called upon for second opinions. Generalists need to support, diagnose, and treat patients across multiple cancer types, are overwhelmed by the amount of work, and can’t possibly keep up with all the new science and guidance for each cancer type. Additionally, generalists often aren’t interested in understanding all the genomic or drug specifics—they just want the results and guidance.​​​​​​​
Responsive Breakpoints (4) + Fluid Layout Resizing
​​​​​​​The Solution
We provided an easy to navigate distillation of the key clinical factors impacting treatment choices while maintaining patient focus with the EHR; allowing oncologists to make informed choices at the point of care based on the latest science (e.g. genomics, clinical trials). Unlike other decision support tools, our Clinical Decision Support solution brings data from the EHR, genomic sequencing, labs, and claims to form a holistic view of all patient data and key insights and surfaces it all in EHR-integrated fashion that with the clinicians, not against them. 

The Pivot
After shadowing users to understand the general process of a oncology department, there were critical technical and workflow constraints that had us change our initial solution approach. The eight oncologists we shadowed always had the EHR full screen. This was a major problem with our free-floating window approach as the user would not have the screen real estate for a secondary window. The users also did very little application switching and spent the vast majority of their time in the EHR.

We decided to go with an always on top, docking application that allowed users to access the solution regardless of their screen set up. This new approach would allow it to work alongside any EHR, such as EPIC or Cerner. The content would float over their primary workspace and be accessible with data to augment their current knowledge of a patient and to support decision-making. We also allowed for flexibility within the delivery system to account for the workflow variations of other customers and users by retaining the original windowed format as an 'undocked' solution.

Value Proposition
Enhance the clinicians current workflow by bringing together and summarize key longitudinal patient data such as current treatments, tumor profiling, genetic and genomic data, comorbidities, imaging data, and lab results—within easy reach by augmenting the weaknesses of large EHR systems.

Utilize the longitudinal data to provide decision support which also factors in: efficacy and related data such as Overall Survival, Complete Response, and Progression Free Survival, toxicity and related data such as risk, life impact, length of treatment, side effects, and cost-related data such as monthly cost, copay, and prior authorization.

Provide AI and ML driven insights to: add to the science of treatment efficacy and comparison, and provide unique insights into outcomes.

Deliver the information and insights at the point of care, in a dual-mode (sidebar dock & floating window), responsive solution that seamlessly integrates with their personal workflow. 

Design Principles
• be Clear: Eliminate ambiguity in what is happening and why.
• be Consistent: Allow users to recognize standards, elements, & paradigms of the system.
• be Efficient: Streamline & optimize workflows. 
• be Accessible: Strive to accommodate the needs of a wide-range of users.
• be Beautiful: Enhance system usability through the use of color, contrast, composition, & motion.
• be Empowering: Give the user agency & allow them to be the hero of their own story.
• be Predictive: Provide insights. Anticipate user needs and make the system feel like magic.

Sidenote: These principles are part of our Precision Design System.

Clinician-specific Design Principles
• Give physicians control & confidence in the science
• Integrated into the current diagnosis & treatment workflow 
• Powered by all available data including genomics
Provide actionable information & insights at their fingertips
• Reduce noise & progressively disclose information as they need it ​​​​​​​
Sidebar docking (always on top of other windows)
App theme examples 
The Team
Our team consisted of product managers, product designers, user researchers, data scientists, cancer researchers, oncology nurses, and engineering leads. We found it critical to have a cross-functional team focused on the product and empowered to solve real-world user problems and meet the objectives of the business.
What We Did
• User & Stakeholder Interviews
• ​​​​​​​Volunteering in a Cancer Center
• Ethnographic Research
• Ideation Exercises
• Information Architecture & Service Design
• Sketching / Wireframing / Prototyping
• Usability, Technical Feasibility, Business Viability, and Customer Value Testing
How We Got There
Our team needed to first identify who our users would be and then start to understand them. Through internal subject matter experts and phone interviews with oncologists and oncology nurses, we created the persona Roberto.

Conducted interviews to learn about the users
We had success recruiting oncologists from within our professional networks. While interviewing, we mostly kept to a two person team—an interview facilitator and a note-taker.

Held proto-persona workshops
As a team we drew out the persona map on the board: Demographics, Behaviors & Actions, Needs & Pain Points, and, finally, a Name & Quote. Using these categories, we drew key insights from our collection of interviews and began to brainstorm realistic user data.

Creation of high fidelity proto-persona
Once the specific data was chosen that we felt best reflected our oncologist, we created the high fidelity persona. This would be used to build empathy for the user within the team. We used Roberto any time we talk or write about features. How would Roberto use this? How does it fit in with the other tools he uses? As we continued to learn more about oncologists, Roberto’s needs and actions were adjusted to be even more realistic.​​​​​​​​​​​​​​
Additional research needs identified
Our product team needed to better understand the environment a solution would be deployed to and the problems it intended to solve. In order to do this, we lobbied for, and secured, two site visits to perform field research.

Sidenote: Our team was already HIPAA-compliant and had signed a BAA with the facility. As an extra precaution, the cancer facility required each team member to complete their OSHA Bloodborne Pathogen certification. This allowed us to shadow oncologists as they treated patients in the exam rooms.

In-person oncology shadowing & contextual interviews
The product team needed to better understand what our intended uses do, and why, and the environment any solution would be deployed to. To this end, we secured the site visit to shadow oncologists, oncology nurses, and cancer patients, and to perform contextual interviews.

The team clipped on our lapel microphones, GoPro cameras, and had field studies clipboards at the ready. We each shadowed an oncologist as they made the rounds, treated patients, and used the software. It allowed us to understand the landscape and obtain insights to how their software played in the actual working circumstances. Where it helped and hindered them. We quickly noticed that the EHR was full screen for the majority of cases and got a firm understanding of how noisy and transient the general working area is. Furthermore we conducted contextual interviews once the providers were finished their overall encounter with the patient. This allowed the oncologists to call out specific pain points while our team of researchers took notes and recorded the audio.

Once we finished with the activities we made sure to share the recordings, transcriptions, and findings with the larger team. These new learnings enabled us to map out their workflow and gave us enough additional information to confidently build the oncologist’s user journey. Through this process, the vision of what problems the product would, and should, solve started to piece together.

Cancer center workshop facilitation
We returned to the facility for series of cross-departmental workshops and design exercises. The user experience team facilitated these sessions over the course of two days. These activities helped us identify general areas where we could provide value and improve their daily work-life.

After the sessions, the team understood what information vital to diagnosis and treatment of cancer patients was typically missing from the EHR. We then brainstormed how we could get that information into their workflow, and bridge any gaps in knowledge for the oncologists. The workshops, along with our earlier learnings, solidified the general direction for the product. We were able to understand the jobs being performed, the pains, and areas of gain. And brainstorm features and functionality to help them perform their jobs with less friction and make their lives easier.
Ideation Exercises
Early Thumbnail Sketches 
Navigation Flow Map 
Conclusion
Through our design process, the team was able to frame what this product will be. We took all the research we had so far such as the personas, user journeys, and field notes, to run internal design exercises to identify the features and form factor we felt were essential to a solution. We sketched, made wireframes, prototyped, and designed in high fidelity to create the initial product.​​​​​​​
Undocked windows
Cancer Informatics Clinical Decision Support
0
12
0
Published:

Cancer Informatics Clinical Decision Support

0
12
0
Published: