Artificial Intelligence for Strengthening Healthcare Systems In Low- and Middle-Income Countries

AI4HEALTH Article Reviews #09

Artificial intelligence has many helpful aspects in healthcare. Obviously, before considering artificial intelligence as a direct helper and benefit provider, it is important to be aware of the limitations here.

This article, which I have read and would like to share with you, was created by compiling articles published between 2009–2021 on artificial intelligence applications used in low- and middle-income countries (LMICs).

That’s why I found this study very informative, which deals with the strengths, weaknesses and perceptions of AI applications with the following 9 important dimensions.

Strengths, Weaknesses and Perceptions of Implemented AIs in LMICs

Reliability of AI tools

✅ Some tools were trained on data outside of their applied contexts, and thus did not fully account for the local disease incidence and treatment options available.

✅ AI models trained on high-income country data may introduce bias into AI outputs, leading to poor performance or, worse, wrong results — which is harmful in a health context and also harmful in establishing AI in healthcare because trust may be broken. it is critical that AI models receive context-specific and updated data on a frequent basis; otherwise, AI models’ performance may worsen over time. This could lead to a downward spiral, as poor performance is likely to lead to poor acceptance of HCWs and a loss of trust in AI-based systems.

✅ It was also observed that the treatment recommendations of IBM Watson for Oncology were in some cases too expensive, not available, considered to be too aggressive or inconvenient for the patient, or locally available alternatives would have been preferred.

Effect on workflows and time to treatment and diagnosis

✅ Although the artificial intelligence applications used close the experience gap of the practitioner and accelerate the decision-making, it has also been observed that it increases the workload in the clinic during the adaptation period.

User-friendliness and compatibility with existing infrastructure

✅ Clinical decision support system was found to require too much information from physicians, which was perceived as too time-consuming in a majority of cases. Lacking integration with existing IT systems also resulted in critical laboratory information not being factored into the AI’s decision-making process.

✅ Poor internet connectivity inside the clinics, crashes of the apps or mistranslations and the overall limited availability of X-ray viewers are other important limitations here.

Trust in AI systems

✅ Distrust in clinical decision support systems, as the basis on which diagnostic or therapeutic decision-making occurred was not sufficiently transparent.

✅ Trust in AI applications has been found to be stronger if a technology and algorithms are understandable and assist users toward their goals.

Cost-savings and improvements in health outcomes

✅ It was seen that only one article dealt with the cost. (AI ‘CAD4TB’ TB-screening tool) It is mentioned in the article that the calculated cost is deemed to be beyond the willingness-to-pay in the Malawian context.

Local adequacy of AI

✅ In this section, topics such as inappropriate treatment suggestions for the area where AI is used, the inadequacy of necessary x-rays and mobile devices are mentioned.

Reported strengths and weaknesses of AI tools.

I think, one of the key points here is to ensure that AI users receive adequate training and to make sure that the practices are indeed compatible with clinical practice in the conditions of the country and region.

Otherwise, the results of artificial intelligence are insufficient to interpret and users avoid extra workload, which turns into a vicious circle that reduces usage.

Therefore, these two topics have a very important place in the design of products to be used in LMICs and produce the results we have discussed in all other topics above.

🌺 Thanks for this valuable paper:
Tadeusz Ciecierski-Holmes, Ritvij Singh, Miriam Axt, Stephan Brenner and Sandra Barteit

📑 Click here for PDF of the paper
Ciecierski-Holmes, T., Singh, R., Axt, M. et al. Artificial intelligence for strengthening healthcare systems in low- and middle-income countries: a systematic scoping review. npj Digit. Med. 5, 162 (2022). https://doi.org/10.1038/s41746-022-00700-y