AI for Good

Chapter 24Detecting Leprosy in Vulnerable Populations

—Yixi Xu and Ann Aerts

Executive Summary

Leprosy is a chronic bacterial infection that damages the nerves, eyes, respiratory tracts, and skin. While the global burden has been reduced by about 99 percent, thanks to the widespread free availability of multidrug therapy, about 200,000 new patients are still diagnosed annually. India, Brazil, Indonesia, and Sub-Saharan Africa see the highest incidence of leprosy. While the disease can be cured, its incubation time can lag between a few months and 20 years. Delayed diagnosis and treatment leave many patients with untreated leprosy and that can lead to irreversible nerve damage, often resulting in disfiguring disabilities. As people with leprosy are often stigmatized and excluded from their communities, it is important to accelerate detection of the disease to enable health providers to offer prompt treatment.

Most forms of leprosy have dermatological features that can be used—with clinical symptoms—to diagnose patients, even in the absence of confirmatory pathology testing. However, there are other diseases that may look similar—such as syphilis, psoriasis, lymphoma, and neurofibromatosis.

Here, we used artificial intelligence models that incorporated photographs of patients' skin lesions and clinical features to help health providers—who might live at a distance from patients, particularly those living in underserved or rural settings—diagnose patients and differentiate those ...