Clinical Robotics
BECTON DICKINSON
My Role
Designing a Safety-Critical Human–Machine Interface for Automated Blood Processing
I led the design of a regulated, safety-critical interface that automated the manual blood-washing process in clinical laboratories. This system reduced human error in handling biohazardous and irreplaceable samples while preserving user trust, procedural rigor, and FDA compliance.
Hardware
User Interface
Problem Statement
Laboratory professionals are required to manually perform the blood washing process—a critical, error-sensitive procedure that involves handling biohazardous and often irreplaceable samples. The manual nature of this process not only consumes valuable time but also introduces a high risk of human error, which can lead to invalid results or the loss of vital diagnostic material.
Boosting Efficiency and Reducing Human Error in Laboratories
To help reduce human error and free up human resources in laboratories, we wanted to build an interface to control a blood-washing device. The interface had to be flexible enough to allow different users to set various parameters and build their own steps to match their protocols. Also, the interface had to be transparent, illustrating the process step-by-step sufficiently so that users could feel confident by the end of the process when they received the blood ready to be analyzed.
Before and After
Challenges
Communicating Trust and Transparency in Automated Systems
Gaining users’ trust that a machine will not replace their work but rather enhance their work and make them feel confident that the result will be flawless requires designing a user interface that gives users control over the machine and shows how it operates. Trust and transparency need to be communicated through the interface. The solution needs to achieve:
Process
Learning About Users
The product owner gave me a clear description of the users and target customers, and also highlighted key stakeholders who are in regular contact with customers. I then set up meetings with these stakeholders, who provided detailed information about the users and their pain points. As the process continued, I independently validated concepts and ideas with some of these users.
Senior Laboratory Scientist
She has been running the flow bench in the immunology laboratory for ten years. She has one full-time senior scientist and two junior scientists supporting her. The juniors prepare most of the samples, and set the machines up for the day, while she and the other senior attend case reviews and review all the files for analysis.
Newly Graduated Laboratory Scientist
He works at a large hospital. He tries to leave on time on a Friday to go home and see his friends. He is based in the hematology department. He fears making mistakes and sending wrong results back to the doctors. For complicated leukemia tests, he will set up the tubes, but a senior scientist will do the analysis with him.
Research
Observational Study
I conducted observational research in laboratory settings, documenting how scientists manually washed samples and where errors or cognitive strain occurred.
Key insights:
- Scientists relied heavily on physical sequencing cues.
- Color coding and spatial ordering were embedded in their mental models.
- Juniors feared making mistakes; seniors feared compliance violations.
- Manual reconciliation between physical steps and documentation created friction.
These insights shaped a hypothesis:
Step 1
Step 2
Step 3
Step 4
Laying Out the Journey of a Blood Tube
The objective of blood washing, also known as RBC (red blood cell) or cell washing, is to remove red blood cells and unnecessary proteins that can distort the data during sample analysis. Thus, our team initially aimed to outline all the necessary steps for users to obtain a clean sample and create an automated process.
I led the cross-functional alignment between hardware engineers and clinical scientists. By mapping the ‘Journey of a Blood Tube,’ I identified that the bottleneck wasn’t mechanical speed, but human cognitive load during setup. I pivoted the design focus to ‘Physical-Digital Syncing’ (the color-coded tube placement), which reduced setup errors by 30%.
The key questions we had to address to develop an interface and ascertain the minimum hardware components needed to process a sample effectively were:
Hardware Workflow
We had to consider our hardware limitations when designing the interface. I graphed the user requirements into sequential steps, which could be seen as building blocks for the device. This helped us brainstorm as a team to establish the parameters the device could handle and that the user could easily manipulate.
Sketches and Wireframes
After the team reached a consensus on what could be addressed through the interface and how it might affect the hardware, I commenced drafting the interface. At first, the graphic concept was unclear, but after engaging in offline conversations with key team members, I began to refine the user interface design.
Rapid Validation in a Clinical Context
Using interactive prototypes, I conducted iterative validation with laboratory scientists.
We tested:
-
Parameter naming conventions
-
Step sequencing logic
-
Error-state visibility
-
Traceability language
Second Iteration Low-Fi Prototype
Each iteration focused on reducing ambiguity and preserving users’ mental models.
Feedback confirmed:
-
Increased confidence in automation
-
Clear understanding of process steps
-
Reduced fear of accidental misconfiguration
Final Solution
Designing for Trust in Automated Systems
Automation risked being perceived as:
-
A threat to expertise
-
A loss of control
-
A compliance liability
To address this, I designed the interface around three principles:
Transparency
Every step of the blood-washing process was visualized in sequence, making the machine’s internal logic observable and predictable.
Control
Users could customize protocols rather than being forced into rigid templates, preserving their professional autonomy.
Traceability
Parameters, reagent names, and step sequencing were clearly labeled and logged to align with regulatory standards.
This reframed the device from “automation replacing human work” to “automation augmenting human precision.”
Interface to add steps and instructions
If the user does not want to use a predefine recipe, she can create her own sequence of steps dragging and dropping steps to obtain a clean sample of blood.
Guiding the user from the physical world to the digital world
Based on the steps the user needs to place physical tubes in a certain order in the robot. The interface provides the order and differentiates the type of tube using a color code as well as alpha-numeric sequences.
Providing users with customization options
Users can select the colors that will identify the position of the tubes in the interface. This feature allows users to mirror the physical world where they operate with the digital environment, which helps them wash samples and add reagents automatically.
Informing the user of errors
Color indicators to highlight errors and missing information.
Customer Reviews and Stats That Support the Hypothesis
Design Impact
Launched as the first automated solution in its category
Reduced procedural friction in blood processing workflows
Lowered risk of manual error in high-stakes lab environments
Maintained regulatory compliance under FDA constraints
Enabled laboratories to scale throughput without sacrificing precision
More importantly, the system demonstrated that automation in healthcare must be designed not just for efficiency — but for trust, safety, and transparency.