Empowering Dermatologists with AI

Overview

Modern health systems face an overwhelming challenge: balancing accessibility, quality of care, cost reduction, and physician burnout. Nowhere is this more apparent than in dermatology, where specialist shortages lead to wait times of six months or longer.

As UX Director at Dermatic Health, I led the design of an AI-powered remote diagnostic platform that streamlined dermatologists’ workflows, increasing efficiency while ensuring clinician oversight and trust in AI-driven support.

The Problem

Long Wait Times for Specialist Access

  • Patients often waited 6+ months for an appointment, delaying diagnosis and treatment.
  • Demand for dermatology services far exceeded available specialists, creating a bottleneck in care.
  • Wait times are a massive drain on revenue, as they are the number one reason for patient leakage, where a patient will find care outside of their usual system.

Physician Burnout & Inefficiency

  • Expert dermatologists spent valuable time on routine cases that could be handled more efficiently, often with alternative providers (e.g. Physician Assistants), who are far less expensive.
  • Physicians spend the majority of their time in non-patient care, mostly inputting information into the EHR, a huge source of frustration.

Financial and Operational Strain

  • Health systems struggled to balance expanding access while maintaining cost efficiency and quality of care. Oftentimes, they are too booked up to even consider expanding access.

The UX Strategy

We focused on augmenting dermatologists’ expertise, not replacing it, through a clinician-first, research-driven UX approach.

1. AI-Powered Remote Diagnostics

  • Designed a system where non-experts (e.g., nurses, general practitioners, and patients) could capture patient images and preliminary data.
  • AI analyzed these images and provided probable condition insights, enabling dermatologists to review cases faster. 
  • The key was to treat AI insights not as a diagnostic aid, but as a workflow aid. We offer up a differential of top probable conditions, the physician can select the one that is correct, and a huge swath of information comes along with that selection that dramatically reduces time spent inputting information into the EHR.

2. Seamless Integration into Existing Workflows

  • Instead of forcing clinicians to learn a new tool, we built the platform directly into their current workflow, ensuring immediate usability and adoption.
  • Dermatologists could review AI-flagged cases within seconds, focusing their time on complex, high-priority patients.

3. Focus on Trust & Explainability

  • AI recommendations were designed to be transparent, showing clear clinical reasoning behind its suggestions.
  • Physicians maintained final decision-making authority, ensuring AI acted as a trusted assistant, not a replacement.

Projected Impact

Revenue Boost from Additional Appointments

  • Average consultation times using the platform would drop from 20 minutes to just 4 minutes, allowing dermatologists to see more patients without compromising care.

Significant Reduction in Patient Wait Times

  • Faster diagnosis and streamlined workflow will mean patients will be seen much sooner, alleviating satisfaction issues and patient leakage.

Increased Revenue & Capacity for Health Systems

  • Expanding specialist access will allow hospitals to increase patient volume and efficiency, driving higher revenue. Additionally, this will allow expert dermatologists to focus on higher level cases, increasing average revenue of each appointment they handled.
  • Expanding access into rural areas, where skin-related ailments are significantly higher than in urban areas.

Securing Investment & Adoption from Leading Hospitals

  • Our design, business strategy, and system performance led to investment from five major university hospitals, including the #1 hospital in the U.S., which committed to adopting the platform.

Key Takeaways

AI should empower, not replace, specialists—by designing AI to support, rather than override, dermatologists’ expertise, we ensured high adoption and trust.


Seamless workflow integration is critical—forcing doctors to learn a new system slows adoption, but embedding AI within their existing tools allowed for immediate efficiency gains with minimal training

Conclusion

By designing an AI-powered remote diagnostic system that worked with, not against, dermatologists, we expanded access, reduced wait times, and improved efficiency—without sacrificing quality of care. This project reshaped the future of dermatology diagnostics, proving that thoughtful UX design can transform AI-powered healthcare.

 

Designs

Skills & Techniques

  • Team Management
  • Figma design system creation
  • Ethnographic User Research
  • Remote user interviews
  • Hi-Fi Prototyping
  • Sketching (Balsamiq)
  • Executive pitches
  • Business Development, customer negotiation
  • Business model development
  • Clinical research
  • AI algorithm collaboration
  • User Testing
  • Product ownership and dev team lead

E-Consults

When physicians need diagnostic help from a specialist, an alternative to referral (asking the patient to schedule an appointment with a different physician) is the E-Consult, which is new to the healthcare landscape. This allows the physician to send the information from the case digitally to a specialist, who can review the information remotely, never having to come face to face with the patient. This saves a huge amount of time for both the patient and the provider.

We built an interface streamlined for E-Consults. If a physician orders an E-Consult, we detect the order and place all associated information into a list of curated E-Consults.

AI Analysis of Photographs

Upon E-Consult request, our system pulls relevant photos from both the physician’s case file and from MyChart, then evaluates them using our AI engine. The AI engine provides a differential diagnosis.

Providing a differential instead of a specific recommendation is extremely crucial here. Instead of serving an exact diagnosis, we can lean on the dermatologist’s expertise. During research, we found that if a photo has enough information for a dermatologist to diagnose, they are typically able to make a determination with seconds. Because of this, the UI is designed around presenting photographs first and foremost, case information second, and then our analysis third.

AI-Assisted Workflow

Because we’re presenting multiple options to the user, and they’ve likely already made a determination of the diagnosis, our AI results serve as a hyper-efficient server of workflow actions.

Along with the diagnosis comes all associated clinical note templates and associated order sets. Instead of having to laboriously dictate and type out repetitive clinical notes and hunt and peck around for associated orders, the user just has to click right from the differential and it builds the entire case in a few seconds.

Fast Results

When the user finishes this E-Consult, we’re expecting a full interaction time of around 2 minutes.

This is down from approximately 10 minutes in current E-Consult workflow, and 15-45 minutes for traditional in patient workflow. Not to mention quality of life improvements for patients, who can expect results in 48 hours, as opposed to weeks and months spent waiting for an in person appointment.

 

These add up to massive cost savings for health systems.