Artificially intelligent medicineMixing AI with telemedicine could allow Nepal to leapfrog in providing healthcare to all
Nepal’s health sector continues to suffer from an equity gap: even when families have access to medical care, it is either unaffordable or of poor quality.
Enter: a combination of artificial intelligence and telemedicine that could allow the country to leapfrog technology and reach people so far deprived of medical treatment.
Over a tenth of Nepalis are forced to live with disease and disability because they cannot afford medical services. In government hospitals where treatment is supposed to be free, quality of care is poor. Many lose their lives simply because they do not get the right treatment in time.
“Creating and using AI models may not sound like top priority for a country where even the most basic healthcare is still inaccessible,” admits Bishesh Khanal, co-founder of the NepAl Applied Mathematics and Informatics Institute (NAAMII). “But those that can benefit from AI are precisely countries like ours.”
NAAMII’s research integrates AI, computing and robotics in healthcare and agriculture. By creating computational models, researchers use machine learning to devise tools to address medical conditions.
With support from IISH (Institute for Implementation of Science and Health) NAAMII is working on an AI-guided screening tool for cervical cancer, which uses smartphone-captured images of the cervix to visually inspect and detect the condition even remotely.
“With visual inspection, gynecology experts can diagnose patients of cervical cancer within minutes,” Khanal explains. “The problem is that Nepal’s socioeconomic and geological conditions make it difficult for such medical expertise to get to places."
With AI tools and telemedicine, caregivers in remote regions can identify disease in consultation with specialists in the city.
Another project NAAMII is working on is an AI-assisted tool that allows a smartphone to act as a microscope to detect diarrhoea parasites in contaminated vegetables, stool, and water.
Read also: Healthcare moving from hospitals to homes, Sonia Awale
AI models that can also detect defects in the intestinal tract, and an algorithm for patients with a muscle disorder called dystonia that can classify what type of the disease a patient has, are some of the other technology NAAMII boasts.
NAAMII’s work in the health sector aims to add to existing care providers so that even in remote regions, advanced medical procedures can be carried out by personnel with less expertise.
AI models are also deployed at the Hospital for Children, Eye, ENT and Rehabilitation Services (CHEERS) in Bhaktapur, under the aegis of the non-governmental BP Eye Foundation. The centre developed a learning-based model integrating telemedicine with AI that allows non-skilled health workers to predict if patients have certain eye, ear, nose, and throat diseases.
Two AI models to predict diabetic retinopathy and glaucoma, both of which are common causes of vision loss, were developed by feeding thousands of photographs to the AI system. These diseases are often overlooked or missed by doctors at clinics, but with both of these models diagnosis was more than 90% accurate.
This disease prediction support is now provided at the Foundation’s outreach clinics across the valley, and also at the health camps organised by CHEERS.
A 52-year-old patient from Lele came to a mobile clinic recently and although diabetic, was not on medication. She thought glasses would correct her vision loss.
An image of her eye, analysed by the AI model at the camp, revealed that she had proliferative diabetic retinopathy. She was referred to a virtual consultation from an ophthalmologist at the CHEERS Hospital in Bhaktapur. Her eyesight was saved.
“Telehealth and AI prediction is a model that can reach specialised diagnosis and treatment more easily to people who do not have to make an expensive trip to a hospital,” says Pranita Upadhyaya an IT professional at CHEERS.
She adds: “Patients in remote villages can now receive virtual consultation with specialists. The AI models further ensure they are treated correctly for the exact condition they have.”
AI also helps community health workers to build capacity and skill, as they can compare and confirm their own diagnosis with ones made by AI and at city hospitals.
But there are also technological hurdles in implementation because of the scarcity of data in rural settings. Upadhyaya explains: “AI models rely on sizeable, varied datasets. These models lack the ability to contextualise like humans, and so when deployed in rural areas where medical records are sparse and datasets are biased, it affects the precision and dependability of AI forecasts.”
To bridge this data gap the 2019 Digital Nepal Framework has provisions for a national digital health mobile application, a centralised telemedicine center, a comprehensive electronic health record, mobile health units, and the provision of e-maternal care.
Read also: An eye for two eyes, Yugottam Koirala
Like with everything else in Nepal, although this blueprint looks impressive on paper, implementation is almost non-existent, even after the government allocated Rs22 billion for it over the past four years.
“Ultimately, how AI enters healthcare is through individualised, specific services,” says Khanal. “There will have to be separate AI tools and products for each niche within healthcare. We will always have to rely on the private sector to keep spurring the development of new AI products. The government’s role should be in formulating and regulating.”
As of now, clear regulatory frameworks governing the use of AI or automated systems in Nepal’s healthcare are practically non-existent. The National Ethical Guidelines for Health Research 2022 sets the standards for research practices in health, but omits any implications of digital technology. The National Science, Technology and Innovation Policy 2019 mentions AI only in passing.
“AI in and of itself cannot be operated without concrete ethical guidelines in place,” cautions Upadhyaya. “When coupled with healthcare, especially in unique contexts like rural Nepal, a whole new array of special ethical considerations arise. Regulations should strike a balance between facilitating innovation and ensuring responsible use of AI technologies.”
Thanks partly to the pandemic, Nepal’s healthcare is shifting into the digital sphere, and this will make deployment of AI in the health system easier.
This shift has been accelerated by public-private partnerships such as Nyaya Health Nepal in Achham, which works with the Ministry of Health and Population to set up Electronic Health Records (EHRs) to remotely coordinate care between doctors and patients.
Says Upadhyaya: “We should move forward by involving government agencies, private sector, academia, and healthcare professionals to drive AI innovation in healthcare while safeguarding patient rights and societal well-being.”
Read also: The Karnali’s hunger emergency, Anita Bhetwal