Leveraging the Power of ML/AI

Praxi Data is collaborating with key personnel from a medical treatment facility to implement innovations to improve “Readiness and Effectiveness.” We sought to significantly improve command agility, data quality, and substantially reduce clerical hours of manual data entry and reconciliation through the use of ML/AI.
‍
What I Did
Facilitated, led and executed 15+ strategic, tactical, and evaluative studies using a range of different user research methods.

Due to the nature of our partnership with our client, I have omitted and obfuscated confidential information in this case study.

RESPONSIBILITIES
 
UX Research
- Project Kick Off
- Stakeholder Interviews
- User Interviews
- User Needs and HMW
- Personas
- Customer Journey Map

TEAM 

Product Design 
Product Management 
Project Management

The Challenge

Delinquent Timecards and Bad Data

Our client currently utilizes the Defense Medical Human Resources System - Internet (DMHRSi) as their timecard submission program. All workers are expected to submit their timecards every two weeks. The data collected from DMHRSi is then used to understand where employees are spending their hours, which effectively impacts multimillion dollar decisions.

However, the program is heavily disdained by its users due to a number of reasons. “Garbage In, Garbage Out” is a common phrase used when it comes to DMHRSi. Compliance is low despite many efforts to stress the importance of the data. In particular, the Medical Treatment Facility (MTF) we are working with currently has a six month backlog of missing and inaccurate timecards. All of these issues make it difficult to properly report performance, assign budgets, allocate resources, and properly align MTFs to patient care.

Interface of the DMHRSi Timecard Submission

Users & Audience

From Top to Bottom

Many roles are impacted by the data that derives from DHMHRSi, from the health care providers who must submit timecards to the Chief Financial Officer who utilizes the data to make decisions.

In light of this, we focused our solution to encompass the needs of multiple users. Our scope was focused on four types of users: health care providers, timecard approvers/timekeepers, DMHRSi Program Analysts, and Chief Financial Officers.

Responsibilities

The Team

This project was a team effort between myself and executives, including the CEO, Head of Project Management, and Head of Product Design.

My role for this project was focused on pairing Human Centered design to identify user needs and balancing it with a solid understanding of the business.

Scope

Understanding the Project

I was placed in charge of the project after the scope had been defined by the company’s executives. To better understand the parameters that had already been identified, I held stakeholder interviews with each executive that had been involved in the process.

From these interviews, I learned that our requirements were to leverage Praxi Data Machine Learning capabilities to:

Build a robust and smart integration between their timekeeping system and electronic health record system.

Provide smart, high fidelity dashboarding and agile reporting for the integration process.

Allow bulk automated actions such as emails or other native messaging modalities.

With my understanding of the basics of the business and the project parameter,I spent my efforts on gathering user data to understand the people who will use the product. For this process, I sought to interview the Chief Financial Officer, DMHRSi Program Analyst, Health Care Providers, timecard approvers, and timekeepers. Each of these roles are impacted by the DMHRSi process and we hoped to gain a holistic view of the entire pipeline.

My main goals for these interviews were to gain an understanding of goals, major tasks, mental models, and opportunities for design to improve the effectiveness of their work.

I worked closely with our client’s point of contact to recruit, set up and schedule the interviews. Over six weeks, I then personally conducted fifteen user interviews.

User Findings

Making Sense of the Data

Upon completing the interviews, I proceeded forward to model the results of my research to better understand behavior patterns, work flows and trends.

Some key insights we learned included:

- How a timecard moves down the pipeline, from beginning to end.
‍
- What prevents providers from submitting accurate timecards on time.
‍
- What common errors are made on timecards.
‍
- How communication works between the different roles.

The insights learned from the interviews allowed me to create user needs and HMW questions. As I was the only designer who conducted user interviews, it was also critical that I was able to build a shared view of the problems, opportunities, and potential next steps with the rest of my team and executives.

I have intentionally omitted confidential data here.

With a better understanding of the users, it was time to create personas to help me determine what the data gathered meant. I created four personas in order to define and design the product by envisioning what users would need most from our product. Each persona was created by focusing on the behavioral patterns that emerged from the data.

These four personas helped me capture user responsibilities, frustrations, and needs. However, it was important for our team to also understand how all these roles impact each other. To help us better create one holistic organization wide vision, we decided to create a timeline and also customer journey maps. These journey maps would also allow us to deepen our understanding of users by considering their emotional needs and thoughts.

Next Steps

Where the Research Takes Us

Through conducting user research, we were able to better identify the issues surrounding the medical center’s ability to obtain the right data, analyze appropriately, and report accurately.

At the time of writing, the project is still currently on going. Praxi Data aims to execute this project by July 2021. Our next steps are focused on creating user and task flows before proceeding on to wireframes.

Our quantitative goals will be to improve DMHRSi time keeping accuracy by over 25% and shorten manual clerical workload hours by over 25%.