AI plays critical role in intestinal worm infection screening among children in Kenya

Experts from Kenya, Sweden and Finland successfully demonstrate a digital microscopy and AI screening technique for worm infections in low-resource settings for the first time.

An Artificial Intelligence (AI) microscopy system has been shown to accurately identify intestinal worm infections in schoolchildren in Kenya, particularly light intensity infections that may be missed by manual microscopy.

A new study – the first to trial the application clinically for worm infection detection in a remote setting with whole-slide images – also demonstrated the viability of implementing the system in low-resource locations.

The study was conducted by a Nordic-East African team from the Karolinska Institutet in Stockholm, Uppsala University, and FIMM (Institute for Molecular Medicine Finland) at the University of Helsinki, and Kinondo Kwetu Hospital in rural parts of Kwale County, Kenya, where there is a high-prevalence of intestinal parasitic worms - soil-transmitted helminths - in children.

During the research, 1,335 school-age children were screened via the deep learning system developed for parasitic worm egg detection, with the results then compared with expert manual microscopy. The findings have now been published in the journal PLOS Neglected Tropical Diseases

 

High diagnostic accuracy

Analysis of digitally scanned stool samples with the deep learning system provided high diagnostic accuracy for detection of three types of parasitic worms. AI was trained to detect Scaris lumbricoides (giant roundworm), Tricuris trichiura (whipworm) and hookworm (Ancylostoma duodenale or Necator americanus) infections.

The AI method accurately detected 76 to 92 % of infections spotted by a trained lab technician, depending on the parasitic species.

Importantly, a substantial number of light intensity infections were missed by manual microscopy but detected by the deep learning system. Notably, in 79 samples (10%) classified as negative for a specific species by manual microscopy, parasitic worm eggs were detected by the deep learning system.

Method is suitable for a resource-limited setting

Principal Investigator (PI) Professor Johan Lundin, MD, PhD, from the Karolinska Institutet in Stockholm and FIMM (University of Helsinki), explained that the application of the AI microscopy technique on stool samples collected from the schoolchildren follows on from the team’s ongoing project deploying similar technology to screening for cervical cancer in women in Kenya.

He said picking up the presence of worms at an early stage, and treating with drugs, has significant health benefits for children.

“Worms are very prevalent in the region,” he said. “In the study, at least 20% of the children had one of the three worms that we looked at.

“We have shown that we can use our testing in a resource-limited setting and get high accuracy. Our method was especially efficient in light intensity infections.”

 

Shortage of experts

Manual microscopy, which can take up to 10 minutes a slide, is the current recommended method of diagnosing soil-transmitted helminths but there is a shortage of experts and access to microscopy equipment and laboratory infrastructure in regions with the highest prevalence.

Professor Lundin described the task for human expert observers doing manual microscopy as like “looking for a needle in a haystack.”

For the AI study, stool samples were collected from the children and prepared according to the Kato-Katz method at a local primary healthcare laboratory and digitised with a portable whole-slide microscopy scanner and uploaded via mobile networks to a cloud environment.

Using the digital samples, a deep learning system was trained and tested for the detection of eggs of the most common soil-transmitted helminths with all AI samples are verified, rapidly, by human observers, he said.

“With AI, once our sample is digitised, it takes just a few second and looks at the entire sample and is able to very accurately find the parasite eggs,” said Professor Lundin.

Additionally, the AI produces a digitally-stored document of the sample which is retained, whereas a human sample dries within hours and becomes difficult to analyse further.

 

AI microscopy can be applied in many areas of interest for global health

Infections caused by soil-transmitted helminths are the most prevalent neglected tropical diseases and result in a major disease burden in low- and middle-income countries, especially in school-aged children.

They can cause chronic infections that result in disability, stigmatization, and significant economic consequences for societies. As the soil-transmitted helminths cause loss of micronutrients, they can cause neurocognitive problems, impaired growth and development and chronic fatigue in affected children. Furthermore, the soil-transmitted helminths are significant causes of morbidity and complications during pregnancy.

Current strategies to reduce morbidity include mass drug administration and interventions related to water, sanitation, and hygiene.

Dr. Nina Linder, Associate Professor at FIMM University of Helsinki and Uppsala University, explained that the AI technique is flexible and applicable in different targets beyond soil-transmitted helminths and cervical cancer.

“The beauty with this technique is that it is multipurpose; we can use it for different purposes with the same hardware with AI trained for different disease or infections", Dr. Andreas Mårtensson, Professor at Uppsala University said.

“But with one scanner you can do a lot - cervical screening and intestinal worms. It can be applied in so many areas of interest for global health."

While new techniques are usually first tested in university hospital settings, he said the team’s AI solutions for worm and cervical cancer screening are being trialled in the rural settings where they will be implemented.

“A great advantage for this project is that we know that it works where it should work,” he added.

 

Significant potential in enhancing screening

Harrison Kaingu, CEO of Kinondo Kwetu Hospital and Health Services, is a Co-PI of the intestinal worm infection screening project, which he says has been positively received by pupils and their families in the poorer Kwale County, which has a poverty rate above the national average.

“The community has poor health seeking behaviour and this contributes to a high disease burden,” he noted. “Neglected tropical diseases are not budgeted for and preventive interventions are poorly funded.”

The project has significant potential across the country and further into Africa in enhancing screening and offering more accurate diagnosis, he said.

The Nordic-East African team say their unique image-based AI technique offers a viable solution for efficient and effective detection of soil-transmitted helminths in children.

“With our point-of-care, minimal infrastructure solution, we have shown that it is possible to use digital microscopy and AI at the very lowest level of the healthcare system,” concluded Professor Lundin.

“Thus, analysis of whole-slide images with image-based AI may provide a future tool for improved detection of soil-transmitted helminths in a primary healthcare setting, which in turn could facilitate monitoring and evaluation of control programs.”

The study was financially supported by The Erling-Persson Foundation, the Swedish Research Council (Vetenskapsrådet), Finska Läkaresällskapet r.f., Medicinska Understödsföreningen Liv och Hälsa r.f. and Wilhelm och Else Stockmanns stiftelse.

 

Original publication: Lundin J, Suutala A, Holmström O, Henriksson S, Valkamo S, Kaingu H, Kinyua F, Muinde M, Lundin M, Diwan V, Mårtensson A, Linder N. Diagnosis of soil-transmitted helminth infections with digital mobile microscopy and artificial intelligence in a resource-limited setting. PLOS neglected tropical diseases. 2024;18(4):e0012041. https://doi.org/10.1371/journal.pntd.0012041

 

More information

Research Director Johan Lundin

Institute for Molecular Medicine Finland FIMM, HiLIFE, University of Helsinki

e-mail: johan.lundin@helsinki.fi