Fighting Cancer Directly From Your SmartphoneBy Miriam Bello | Fri, 05/15/2020 - 17:01
Q: What areas of opportunity did you identify to participate in this market?
A: We started this project in 2017, when one of my closest friends, was diagnosed with skin cancer. Fortunately, he was able to pay the high costs of prevention. However, this is not the situation surrounding millions of Mexicans and people around the world.
We live in a country where the minimum wage is roughly US$5 per day and 96 percent of the population has a monthly income below UDS$750. The average Mexican family spends US$650 a month on household expenses and the private dermatology consultancy costs between US$50 and US$100. It is virtually impossible for people to manage a regular checkup just to see if everything is alright.
To put things into perspective, in 2018 there were only 1,784 certified dermatologists across all of Mexico. This is a country with a 123 million population. Which means each dermatologist would have to see about 69 thousand patients per year. At 15 minutes per consultancy, eight hours of work per day, five days a week, 52 weeks per year; dermatologists are roughly capable of attending about eight thousand patients per year, leaving a 60 thousand deficit per dermatologist.
Q: What alliances are you making with dermatologists and how are you convincing them to participate with Dermapp?
A: When we started the project, we solicited advice from dermatologists because we wanted to create an appropriate platform right from the beginning. We are always looking for feedback and different points of view to ensure we are offering the right solution to the patient. The product is not targeted for dermatologist use, however; it is for patients. If a doctor’s visit is level one of attention, this would be level zero. The goal is for this technology to provide enough information to serve as a preventive method. It is a monitoring tool for patients, so when they detect a red flag, they can go to a specialist to receive a proper diagnosis.
We recently started working with a version for primary-care doctors. In Mexico, only 58 percent of the population has access to the medical devices needed for cancer diagnosis, so they can use this tool as a first step. We intend for this to be a support tool, not a diagnostic tool. General doctors can use it to start the diagnostic process and later decide where to canalize the patient.
Q: How have you introduced this technology to medical staff and patients?
A: The medical sector is skeptical of new technologies, which pushed us to improve the credibility of our project. We focused on having the academic background and scientific support that would give Dermapp and our company enough credibility to gain trust from doctors and patients.
Our first approach with patients has been positive. Today, patients tend to look at their symptoms online rather than going to a doctor. This has caused problems for the healthcare sector because it leads to self-diagnosis, which could endanger the patient. As a result, some doctors believe our platform will serve as a self-diagnostic app that allows patients to skip checkups. Right now, we are focusing on gathering enough scientific evidence to counter these assumptions and to increase the reliability of the platform.
Q: How precise is Dermapp compared to a more traditional approach to initial diagnosis?
A: The app works with two algorithms, one for general doctors and the other for patients. The latter is exclusively focused on melanoma. Technology-wise, we used a machine learning system to develop an algorithm with 75 percent accuracy and a 0.8 area under the exactitude curve, which makes it a good method of diagnosis. Nonetheless, we do not intend Dermapp as a diagnostic tool but to help with a proper clinical diagnosis. Validation was done through traditional standards of the artificial intelligence industry in which images of common moles, dysplastic nevus and melanoma were stored in a database of approximately 6,000 images. The algorithm was based on the comparison between our database and other images to detect benign and malignant lesions.
The second algorithm is a bit more complex because we use deep learning and artificial vision that focuses on image processing. This allowed us to develop an algorithm that compares pictures against a database of 25,000 images of different types of lesions like melanoma, basal cell cancer, spinocellular cancer, actinic keratosis, common nevus, benign basal lesions, seborrheic keratosis and dermatofibroma. We reached an 84 percent accuracy rate on malignant lesion detection. The difference is that this algorithm requites dermatoscopic images, which forced us to develop a low-cost dermatoscope adapter for phones that allows observation of structures under the surface of the skin to help the dermatologist and the algorithm to obtain more information for the diagnosis. The cost is way below what is available in the market. Commercial dermatoscopes cost around MX$9,000 (US$508), while we are offering ours at MX$2,400 (US$125.8). Our hope is to lower this number to MX$600 (US$31.4) when we start mass production. Our long-term goal is to make it possible for any person to purchase this dermatoscope but at the moment it is only targeted to general doctors.
Q: Beyond dermatologists, what other alliances are necessary to increase the use of Dermapp?
A: We approached ISSSTE in Nuevo Leon to conduct a pilot test at two clinics where 19 general doctors would use the app and the dermatoscope on approximately 14,000 people per month. We would like the proposal to be directed to the public sector because that is where resources are limited. ISSSTE in Nuevo Leon has around half a million beneficiaries and only three dermatologists based in Monterrey, which means that outside the metropolitan area, people do not have access to dermatological treatments. There is a telemedicine system through which patients in rural areas can receive an appointment but the waiting list is from one to two months. Those months can be decisive in relation to skin cancer. We want to empower general doctors to detect high-risk cases so institutions can prioritize their resources properly.
Q: How will artificial intelligence change the health sector nationally?
A: Artificial intelligence can optimize and ease the use of technology, which leads to a more dignified healthcare system. Coverage can be complicated and in Mexico we are not quite there yet in terms of complete healthcare coverage, which means that access to specialized services is limited. Technology allows us to change this reality and benefit the people through access to traditional healthcare.
Big Data also changes the panorama. Collecting data on a large scale leads to predictability and patterns in clinical cases, which is beneficial for preventive medicine. Last year, the FDA approved the first medical device based on artificial intelligence and focused on diabetic dermopathy, which provides a diagnosis without the need of a doctor’s approval. This is where we would like to get to eventually. The benefits that this would bring to rural communities in Mexico, for example, would really change healthcare coverage and the overall healthcare system.
Q: How can public and private institutions work together to reduce the prevalence of skin cancer?
A: Technology can empower primary care doctors to detect this disease as soon as possible to reduce risks and costs. In its early stages, the treatment is not that invasive and in the private sector costs around MX$4,000 (US$209.5), which is minimal compared to what chemotherapy costs.
Q: What are your priorities for 2020 and how will you grow your user base over the year?
A: We would like to continue validating our product and working on R&D. We are also working to increase our image database so we can create a more exact and robust artificial intelligence algorithm. The app is already available and we are monitoring people’s response, seeking feedback and ways to improve our system and facilitate its use. We have already identified areas of opportunity, such as improving the algorithms for image position that are used to detect lesions. In 2020, our goal is to improve the app and seek validation in a more realistic environment, where we get to compare its diagnostic performance with that of a general doctor or dermatologist. The impact of those results would raise the app’s credibility and trust.
Monterrey HealthTech Solutions is the Mexican company that created Dermapp, an AI-driven mobile app for the remote assessment of skin cancer risk. The app analyzes user-taken photos of skin moles to predict the potential for developing skin cancer with 75 percent accuracy.