Similar Patients

Syapse Oncology Web Capability

With Syapse, oncologists can view treatment data for patients across the entire health system, based on tumor type, stage, genomic variants, demographics, and other clinical parameters. This population-wide view enables oncologists to recommend a targeted drug based on institutional prescribing patterns, reducing variation in care.

My Role & Team

Lead Product Designer

I worked on Similar Patients from conception to implementation - and continue refine, build and improve the feature today.  This includes:

  • User Research 
  • Iterative Design
  • Translation to Product Development

Team

I worked closely with the product manager from the minimum viable version of the feature to continuous improvements on the feature. Together, we collaborated with stakeholders and engineering teams to define, build and improve Similar Patients.

 
 

Using Real-World Data to Drive Clinical Decisions

CHALLENGE

In oncology today, patient data is siloed and inaccessible. Within a healthcare system, critical information about a cancer patient’s family history, disease status, genomics, and treatments are locked inside difficult-to-access health IT systems, such as electronic medical records.  

Hence, physicians base their treatment decisions on clinical trial publications and standard of care guidelines - information that is slow to update and difficult to navigate in order to identify what information is relevant for a given patient.

Syapse Precision Medicine Platform reaches into legacy systems and aggregates health history, medical records, cancer status, labs, radiology, pathology, molecular, genomic, and treatment data.

With the integrated data, our goal was to help physicians easily access and learn from aggregate clinical, genomics and outcomes data.

Constraints

UX Team of One + Aggressive Timeline 

When the idea of Similar Patients was born in 2015, I was the only designer in Syapse juggling this project along with other product features. I worked in a lean approach to design as time and resources allowed for the first version of Similar Patients. This meant owning user research, design and user testing for the Similar Patients - otherwise no one will do it. 

For the 2017 latest version of Similar Patients, I had the joy of growing the design team and work with another UX designer to take Similar Patients to the next level.

Varying Data Models and Content

Each health system has their own data model. Its important for the engineering team to be able to successfully map each health system’s data model to the Syapse model. 

Data Quality

It's important to understand if the content can be standardized and used across health systems - and identify if standard libraries can be utilized so this content can be easily accessible and understandable in the clinical setting. Clinical staff must also trust how that data is collected within their health system.

Process

What does it mean for patients to be similar?

RESEARCH

Through informational interviews with subject matter experts and physicians, we collected all the various questions physicians may ask given particular patient situations.

We learned that the level of genomics knowledge of the physician influences how a physician might define “similar patients”:

  • From the traditional disease-based line of questioning “For lung cancer patients with stage 4 cancer…”

  • To genomics-based line of questioning - “My patient has EGFR and TP53 mutations, are there other patients who have this molecular profile?”

Furthermore, physicians had a wide range of information that they would want to know if they had access to data for a given patient cohort. While there were common themes to the information, physicians were skeptical that any of this information could be feasibly displayed either due to data quality or lack of available data.

From Vision to Minimum Viable Product (MVP)

Design

Based on our interviews, I defined the following design goals to help us focus on the user experience:

  • Identify core data elements that can be used across customers

  • Make it easy and flexible for physicians to explore population data

  • Build trust in the data

Internal Collaboration Sessions

Results of a collaboration session 

Syapse has a range of employees who are either subject matter experts in genomics, oncology and know our customers well. To get a variety of perspectives on this project, I gathered our internal experts through a series of collaborative sessions. These included:  

  • Discovery: where we explored the top research questions

  • Brainstorming: Brainstormed possible solutions using research data

  • Feedback: Rough designs and prototypes

These sessions served as a great way to involve people across all departments at Syapse, socialize and build ownership around the feature and get a diversity of ideas / perspectives.

Shaping Similar Patients  

designs to facilitate data exploration

Research showed that physicians are interested in exploring clinical data from various perspectives - it could be from a disease-focused question, genomics focused, demographics or any combination of the above. I explored designs for a visual query tool to make it easy for physicians to explore the data. The long term goal from a product standpoint is to be able to add other categories of clinical data for exploration over time. 

 
 

Exploring Data Visualizations

For V1 of Similar Patients, I explored various visualizations that can help physicians understand the patient population in terms of their genomics and disease. The visualizations needed to be flexible to accommodate the various lines of questioning physicians can put into the application.  

 

Simple animation highlighting a hover effect

 

User Feedback Sessions

I organized user feedback sessions and prepared click-through prototypes for several rounds of user feedback sessions in order to validate the clinical scenarios and how a dashboard tool can be a decision aid for the physician. These sessions helped us identify if the right information elements are applicable and helped us narrow down clinical questions physicians are asking.

 
 

Build, Deploy, Learn and Iterate

Improving Similar Patients Over Time

Using real-world data for the clinical setting had never been possible before for our customers. So it was important for Syapse to partner with customers to understand how they use it and identify further opportunities to improve and enhance the tool to provide value for physicians at the point of care. Once physicians understood its capabilities, they are able to point us to other data that they might be interested in or other ways to leverage it within their workflow. This partnership enabled us to have regular forums of feedback with our customers, with the product team visiting various medical centers for on-site observations and holding regular user feedback sessions over time. 

 
 

Similar Patients

Result

Similar Patients is a visual query tool that allows physicians and clinical staff to view treatment data for patients across the entire health system, based on tumor type, stage, genomic variants, demographics, and other clinical parameters. This population-wide view enables oncologists to recommend a targeted drug based on institutional prescribing patterns, reducing variation in care.

 
 

Easily define a patient cohort based on any combination of clinical and molecular characteristics

Physicians have the flexibility to quickly add patient characteristics to define their patient population. The physician can add, remove and edit their query - while getting real-time results for each change to their query. Or the physician can select specific patient and Syapse will pre-populate the query builder and identify similar patients.  Each component was carefully designed to make it easy for physicians to modify structured clinical and molecular data. 

 
 

View Results and Learn from Similar Patients

Physicians can view real-time results as they enter clinical and molecular characteristics. The query is made against all patients in the organization and returns aggregated, anonymized statistics. Physicians can understand the other clinical and genomic make-up, such as the disease and stage of the patient’s cancer, as well as the top reported mutations by gene and alteration. Furthermore, physicians can learn top treatments prescribed to similar patients across the organization.

 
 

Inspiring Data Sharing Across Health Systems

The MVP version of this feature helped customers recognize the opportunity to join together to advance cancer care through sharing of cancer genomics data. This led to the evolution to Shared Similar Patients, where physicians can view patient data across organizations and use these treatment insights to influence patient care.

 
 

Customer Stories & Press

“This consortium exists because we all arrived at the same important conclusion: we need to collaborate across health systems to cure cancer.
”
— Dr. Lincoln Nadauld, Intermountain Precision Genomics
“By aggregating all of our real patient experiences, we will rapidly expand our ability to learn how to choose the best targeted treatments for our cancer patients based on the molecular profile of their tumor and our informatics based research.”
— Dr. Jim Ford, Stanford Cancer Institute