Zeroing in on the last mile
More data, more precise
Overcoming challenges
Patient with a leprosy lesion marked by an expert. This data is later fed into an AI algorithm.
Science
00

Tackling the world’s oldest disease with the latest technology

According to the World Health Organization (WHO), more than 200 000 people worldwide were newly diagnosed with leprosy in 2018. This is not only a very high figure, but also a long way off the WHO goal to eliminate the disease altogether. To support the WHO in its quest, the Novartis Foundation is collaborating with Microsoft and is using artificial intelligence (AI) in the fight against leprosy.

Text by Patrick Tschan, photos by Alexander Kumar

scroll-down
Home
en
de
zh
jp
Share
Share icon
content-image
Enter fullscreen

Although leprosy can be treated, lack of healthcare infrastructure is a key impediment to eliminate the disease.

arrow-rightZeroing in on the last mile
arrow-rightMore data, more precise
arrow-rightOvercoming challenges

Published on 23/11/2020

In 2018, the road movie Yommeddine was selected to compete for the Palme d’Or at the Cannes Film Festival, telling the story of a cured leprosy patient and his difficult reintegration into society. The film threw a rare spotlight on the disease and showcased the plight of one of the hundreds of thousands of people worldwide who are suffering from leprosy, which – although curable – has defied eradi­cation for decades.

Leprosy has been treatable with multidrug therapy, or MDT, since the 1970s (see box on page 48). Despite this possibility, the disease remains a huge global health challenge as the medical therapy constitutes only one element in the fight against the world’s oldest illness. Other challenges are the diagnosis of the disease and the early prevention of new infections. 

The number of new patients diagnosed with leprosy remained persistent at about 200 000 to 250 000 per year in the past decade. This is partly due to the fact that leprosy has an incubation period of more than 20 years and often occurs in hard-to-reach communities with poor access to healthcare. Hence, untreated patients can easily spread the disease, especially in the early phases. 

content-image
Enter fullscreen

A photographer is taking high-resolution pictures of a leprosy patient.

Ze­ro­ing in on the last mile

This so-called last mile in the fight against leprosy has been a key focus of the Novartis Foundation. Over the past 20 years, it has established several programs to help stop the transmission of leprosy. 

“Besides addressing the diagnosis and treatment for leprosy patients, screening and post-exposure prophylaxis for people who have contact with newly diagnosed patients, our programs pay special attention to early detection, which enables us to provide swift intervention,” says Leprosy Project Manager Gani Zaahira.

One such program in the Philippines focuses on supporting community health workers with smartphones to send photographs of suspicious-looking skin lesions that can be the first symptoms of leprosy, and send those images to reference specialists. The doctors who are part of this so-called LEARNS (Leprosy Alert Response Network and Surveillance System) program provide diagnosis and initiate the necessary treatment. To date, more than 5000 healthcare providers have been trained to use this tool, allowing them to accelerate leprosy diagnosis significantly and to correctly identify 75 percent of suspect lesions as leprosy. 

Artificial intelligence for the last mile

Given the success of LEARNS, the Novartis Foundation is now stepping up its efforts in the digital domain and has started working with Microsoft to create a completely new digital diagnostic accelerator mechanism. 

Together, they are developing a tool using AI to detect leprosy with the help of image analysis. “This may enable us to eliminate the world’s oldest known disease with the newest technology we have available,” says Ann Aerts, Head of Novartis Foundation. “The use of digital technology can truly support us to accelerate the last miles and make leprosy history once and for all.”

The new diagnosis accelerator is based on images of skin lesions. “By October 2019, we had captured over 800 images from 170 leprosy patients. Extremely high-resolution images of arms, legs or backs with skin changes that are mostly leprosy,” explains Johannes Boch, Stakeholder Engagement Lead at the Novartis Foundation, who manages the AI4Leprosy project for Novartis. 

“Some images capture up to 150 leprosy skin lesions in one shot.” Each lesion is labeled by dermatologists and classified as either leprosy or not leprosy. The result is a visual description of the affected parts of the skin. The image data together with other patient characteristics are loaded into the AI algorithm developed by Microsoft. Based on this data, the algorithm “learns” to distinguish healthy from affected skin.

content-image
Enter fullscreen

Hard to reach regions are another major obstacle...

More data, more pre­cise

The current data comes mainly from Brazil, which has one of the highest number of leprosy patients worldwide. The Novartis Foundation is partnering with the Oswaldo Cruz Foundation (Fiocruz) to collect the images and combine them with data from leprosy patients: Their medical staff diagnoses the disease and classifies the symptoms.

The result of this process will be a proof-of-concept that AI can be applied to recognize leprosy skin lesions. This could then hopefully be used on a mobile phone. The application could potentially allow any user to photograph a skin lesion, send it to the cloud and receive from the AI algorithm a probability of the lesion being leprosy. This could provide decision support to inform further diagnostic steps.

“At present, the AI algorithm is being trained to correctly differentiate between lesions and normal skin and to recognize leprosy. After the proof-of-concept, the objective is to make the algorithm more robust through the inclusion of additional pictures and data from India. This would allow us to start including different quality images and further train the algorithm,” Johannes Boch says.

Gangadhar Sunkara, who leads the team in India, holds similar aspirations: “India still has a considerable leprosy burden. Not only stigma and access to treatment are still critical but also early diagnosis is a challenge. The use of digital technology and the potential we have through the development of an AI-powered tool to accelerate diagnosis can help to address this burden early enough so that patients do not have to suffer from the devastating effects of an untreated disease.”

The AI4Leprosy team hopes to enable a correct diagnosis of leprosy for the majority of the cases.

Thumbnail
Play iconPlay icon
More than words: fighting leprosy with artificial intelligence
content-image
Enter fullscreen

 ...to make leprosy history.

Over­co­m­ing chal­len­ges

There are many difficulties in getting quality data. “Challenges such as the availability of trained medical staff who can identify and describe leprosy, or achieving high-quality data sets for the proof-of-concept phase, and high-performance computers with strong internet to upload high-resolution images, are repeatedly delaying progress,” Johannes Boch explains, listing some of the real-world conditions they confront.

But Boch is confident the initiative will be successful despite these hurdles: First, Brazil has more mobile phones than inhabitants. Second, the country is equipped with a well-developed mobile network, even in rural areas. “There is reason to hope that with simple information and education in the communities, the solution can be beneficial in accelerating diagnosis and supporting clinical decision-making,” says Johannes Boch. For example, remote diagnosis can help identify leprosy at an early stage and allow patients to be treated before they develop irreversible neurological damage or further spread the disease.

The goal to eliminate leprosy, as set out in the London Declaration on Neglected Tropical Diseases, seems very ambitious given that new case detection has stagnated over the past 10 years and more than 200 000 new leprosy diagnoses were still registered last year. A user-friendly AI-powered accelerator of leprosy diagnosis could certainly play a key role in the search for undiagnosed patients. If implemented across endemic countries, it could make a significant contribution to further reducing the spread of leprosy and making the lives of patients – unlike the plight shown in the movie Yommeddine – much easier.

icon

Multidrug therapy, or MDT, consists of three antibiotics, including two compounds developed by Novartis predecessor companies. The therapy has had a great impact, helping reduce the number of leprosy cases by more than half from 5.3 million in 1985 to 2.2 million in 1993. In the past 20 years alone, more than 16 million leprosy patients have been treated with MDT.

Novartis played a major role in this success story. Since the year 2000, the company has been providing MDT free of charge to leprosy patients via the World Health Organization, having donated more than 56 million blister packs valued at above 90 million US dollars. These efforts helped treat over 6 million patients worldwide. In 2015, as part of its commitment to the London Declaration on Neglected Tropical Diseases, the company announced the extension of its program through to 2020.

You may also be interested in the following:
novartis logo
Back to the main page
Patrice Matchaba: Group Head of Global Health & Corporate Responsibility for Novartis.
29.11.2020
«We have to include the developing world in the innovation cycle»

Developing access.

Lutz Hegemann, Chief Operation Officer Global Health - Lutz oversees the global health operations of Novartis. This includes the group’s overall access to healthcare operations, including business in sub-Saharan Africa, where Novartis is implementing a new model with a primary focus on patient reach.
15.11.2020
Access makers

People Power.

Home
en
de
zh
jp
Share
Share icon