A multi-app healthcare platform powered by ML that offers skin condition assessments while generating leads and improving diagnostic data collection

Innowise has developed an AI-powered app that uses deep learning and image recognition to quickly assess skin conditions, giving fast, preliminary diagnoses based on uploaded photos.

Customer

Industry

Healthcare

Region

Central Asia

Client since

2024

Our client, a leading dermatological clinic network in Central Asia with over 10 years of expertise, serves more than 1,000 patients daily across six countries. They focus on areas like allergology, phlebology, dermatological surgery, and more. Their approach blends patient-centered care with advanced diagnostic tools and the expertise of top specialists. This combination enables them to offer services ranging from managing chronic skin conditions to providing aesthetic improvements. Known for their patient-centric approach (NPS > 9) and catering to a clientele that includes 12% high-net-worth individuals, they sought a solution to strengthen their position as innovators in the region.

Detailed information about the client cannot be disclosed under the provisions of the NDA.

Challenge

Develop an AI-powered diagnostic app that helps establish market leadership and attract high-value patients

With increasing competition in the region, the client recognized the potential of AI not just for improving diagnostics, but as a powerful marketing tool. They wanted to attract new patients, particularly in the high-net-worth segment, and position themselves as technology leaders in the Central Asian healthcare market.

For this purpose, the client decided to develop a ML-powered mobile app to automate the preliminary diagnostics of skin conditions. A key challenge here was the need to acquire and maintain high-quality image data for training and validating an ML model, aiming for ambitious accuracy targets while acknowledging the limitations posed by variable image quality. Without an internal development team to bring this vision to life, they reached out to Innowise for software development services.

Solution

An AI-driven platform integrating mobile apps and a web admin panel

Innowise developed a comprehensive platform comprising two interconnected mobile applications and a web-based administration panel, all powered by a custom-modified DINOv2 model using transfer learning with Convolutional Neural Networks (CNNs).

Patient app (iOS and Android): This app serves as an advanced marketing tool, offering users a free, ML-powered preliminary skin assessment. This innovative approach provides instant assessments for 30 skin conditions, acting as a lead generation tool for the clinic network. The app’s user-friendly design and personalized recommendations encourage users to book consultations at the client’s clinics.

Physician photo collection app (iOS and Android): This app allows clinic staff to securely capture and upload high-quality images of various skin conditions, directly contributing to the ongoing training and refinement of the DINOv2 model. This continuous feedback loop ensures the AI remains accurate and up-to-date. The app also includes a reporting system for tracking photo statistics and diagnosed conditions, providing valuable data for analysis and improvement.

Web-based administration panel: This panel provides clinic administrators with comprehensive tools to manage diagnoses, configure treatments and medications by country, review AI-generated assessments, analyze app usage data, and generate reports. This centralized system streamlines operations and provides valuable insights into patient demographics and trends.

The entire platform is built on a scalable and secure AWS cloud infrastructure, ensuring data privacy and reliable performance. The initial dataset for the DINOv2 model was provided by the client and is continuously augmented by images collected through the physician app.

How does the skin scanner app work

The skin scanner app is designed for ease of use, guiding users through a simple process to receive a preliminary assessment. From body part selection to personalized clinic recommendations, the app provides a seamless user experience. Here’s how it works:

Body part selection: When users open the app, the first step is selecting the part of the body where the skin condition is located. This helps the app narrow down the possible conditions that could relate to that specific area.
Image upload: Users can either take a photo of their skin condition or upload one from their gallery.
Questionnaire: Once the photo is uploaded, users answer a short three-question quiz. These questions help add some context for the machine learning analysis, like symptoms or any relevant medical history.
Image analysis and diagnosis: After the photo has been submitted, the app provides three possible diagnoses, each with a probability score. For example, it might show acne (80%), dermatitis (15%), and psoriasis (5%). .
Detailed condition information: Users can tap on any diagnosis to get more detailed info about the condition, including a description, treatment options, and recommended medications. This info is regularly updated through the admin panel to keep everything up-to-date.
Geolocation-based clinic recommendations: The app uses geolocation to provide users with a personalized list of nearby clinics where they can get treated for their conditions. Each clinic comes with all the contact info and exact locations on an interactive map — making it easy for patients to connect with healthcare professionals. If there aren’t any suitable clinics in the user’s city, the app suggests alternatives in nearby cities or regions.
User registration and profile management: The app offers users two options: guest mode and registered mode. In guest mode, users can get quick diagnostics without setting up an account. Registered users, on the other hand, unlock extra features like a personalized profile where they can keep track of their diagnosis history, save photos, and get more detailed insights based on their past interactions.
In-app advertising: We helped the client add non-intrusive banner ads to the app, strategically placing them at the top or bottom of the screen to create an extra revenue stream.

Technologies

Mobile

Flutter

Frontend

Angular

Backend

Python, FastAPI

Machine learning

DINOv2, AWS SageMaker

Security

TLS, AES-256 Encryption, MFA

VCS

Git, GitHub

Cloud

AWS

Process

A phased approach ensured smooth execution, from discovery (photo collection app demo and workflow design) to implementation (mobile development, model training, and infrastructure setup) and finally, continued operation and support (ongoing model refinement, knowledge transfer, and dedicated support).

Team

Project Manager

Business Analyst

Angular Developers

UX/UI Designer

Python Engineers

Flutter Developers

ML Developers

QA Engineer

Results

A successful AI-powered skin diagnostic platform achieving rapid user growth and high diagnostic accuracy while generating expansion opportunities

We’ve developed an ML-powered mobile app that provides users with a quick and secure way to assess their skin conditions. Within the first three months, the cross-platform app has gained 5,000 new users, helping the client carve out a strong presence in a competitive market. Alongside this, we created a photo-collection app to train and fine-tune the ML model, which now achieves 80% accuracy across 30 dermatological diagnoses.

Our team also built a web-based admin panel that lets clinic admins manage content, track usage, and keep all the data up to date easily.

Looking ahead, the client entrusted our team to implement subscription options and build API access to the model for a network of partner clinics. We’re also working on improving the current features to keep the app as effective and user-friendly as possible.

Project duration

February 2024 - Ongoing

5,000

new users in the first three months

80%

ML model accuracy achieved

Related cases

Contact us!

Book a call or fill out the form below and we’ll get back to you once we’ve processed your request.

What happens next?

Having received and processed your request, we will get back to you shortly to detail your project needs and sign an NDA to ensure the confidentiality of information.

After examining requirements, our analysts and developers devise a project proposal with the scope of works, team size, time, and cost estimates.

We arrange a meeting with you to discuss the offer and come to an agreement.

We sign a contract and start working on your project as quickly as possible.