“Right now, AI is used where it is least needed,” he says. “We’ve flipped that perspective and brought it to environments that lack the necessary expertise and resources.”
Professor Lundin is currently collaborating with Kenya and Tanzania, where the need for pathologists is desperate.
“In Sweden we think we have problems with our 30 to 40 pathologists per million population, but in sub-Saharan Africa there are less than one per million. »
He hopes to use AI and digital technology to expand access to image-based diagnostics. The components to digitize the samples come from the mobile phone industry and the data can easily be transmitted elsewhere.
“The person who evaluates the microscope images could be in another city, or even in another country,” he says.
Screening of 400 million women
In another project, researchers are working on cervical cancer. In the Nordic region, the disease is prevented through screening and vaccination, but in many low-income countries it is the most common cause of cancer death among women.
The World Health Organization (WHO) has set a goal of having 70% of women of screening age tested by 2030.
“To achieve this, an additional 400 million women need to be screened worldwide, which will be difficult to achieve without automated methods. »
Researchers worked with staff to screen more than 3,000 women at Kinondo Hospital in Kenya, and another 600 women in primary care in Tanzania. A nurse takes swabs from the ectocervix, after which the cell sample is placed on a microscope slide, prepared and scanned. The AI then analyzes the sample and an experienced pathologist verifies the AI response remotely.
Here, the researchers were able to show that the AI was no less accurate than an expert. They also analyzed samples for the presence of the virus that causes cervical cancer. In many countries, this is the primary means of screening for disease, and women who test positive must provide additional tissue samples.
“But in regions where 25 to 30 percent of women test positive, this is difficult to implement – in which case AI sample analysis can be a ‘middleman’ that relieves some of the pressure on the system,” says Professor Lundin.
Finds pests in seconds
Soil-spread intestinal parasites are a group of “neglected diseases,” although some 1.5 billion people worldwide carry them and 20 to 30 percent of children in some areas are actually infected. Infections are treated with deworming preparations taken in a single pill form, which has led to mass treatment. However, repeated treatment can lead to drug resistance.
“Our method allows doctors to only treat those who are actually reinfected after the first treatment,” he explains.
Researchers tested fecal samples from 2,500 schoolchildren. When a microscopist analyzes the samples for worm eggs, it takes 10 to 15 minutes per slide; for AI, it only takes a few seconds.
“And in more than ten percent of cases, the AI found eggs that the human expert had missed,” says Professor Lundin. “There may only be one or two eggs per slide, so it’s like looking for a needle in a haystack.”
One challenge is sample preparation, as reagents can vary between production batches and laboratories.
“AI is like a student,” he continues. “If the samples look different from one occasion to the next, the AI is less effective.”
For this reason, the AI model must be adapted to local conditions through standardized routines and quality checks, as described by the researchers in The British Medical Journal in 2025.
“When AI is introduced locally, the model must be adjusted manually ensuring the quality of the first 50 to 100 samples,” explains Professor Lundin.
But in a few years, that could be a thing of the past.
“Maybe we won’t need to dye preparations like we do today. We had to do it to help the human eye, but AI might be able to detect anomalies without it.”
Professor Lundin highlights that AI has the potential to reduce global inequality if used responsibly, in a spirit of trust and education.
“We have seen great support for these developments in low-resource environments, especially due to the lack of experts. AI cannot be a high technology reserved for rich countries, because low-resource countries have the greatest need for it.”
Support for personalized medication
While AI is often used to detect abnormalities in different types of images, it can also analyze data to identify medically relevant patterns, such as a subgroup of patients standing out for their particularly good – or poor – response to a certain drug.
Helga Westerlindprofessor of epidemiology at Karolinska Institutet Department of Medicine in Solna, develops AI models to search for such new patterns in a vast database.
In one project, she focuses on patients with rheumatoid arthritis, an autoimmune disease. When diagnosed, most patients are given the immunosuppressant methotrexate, although a third must stop taking the drug within a year due to ineffectiveness or side effects, and other treatment options are available.