The Case for AI in Global Health

Cooper/Smith
4 min readJun 29

--

Photo credit: AP

We’ve all read the warning stories about AI: sentient robots, deep fakes, disappearing jobs. If the risk-benefit calculation for use and regulation of AI focuses on existential risk of human annihilation vs. the trivial benefit of creating a computer-generated pop song or digital image, then the obvious action ought to be to minimize the influence of AI in our lives.

But this framing fails to consider the massive potential upside for AI in the health sector, particularly in low-income countries. Over-focusing on the risks overshadows the life-saving benefits of AI and global public health, especially in health care for people in low- and middle-income countries. Failing to recognize these benefits could result in clumsy or poorly-considered regulations and policies that would worsen our lives, not improve them.

Thanks to AI, we are on the cusp of a transformative, life-altering revolution in global public health. Deploying AI in low and middle-income countries leapfrogs existing medical and public health infrastructure, adds essential capacity to countries with a dearth of doctors, public health workers, and aid specialists, and accelerates vaccine development that will dramatically improve health outcomes.

For example: poor countries tend to have a limited number of doctors, nurses, and other healthcare cadres — a key contributor to poor health outcomes. It takes a long time to train and add healthcare workers to the system, so this situation is not likely to improve for a long time. Poor countries are also more likely to have physical access challenges to healthcare, with many people having to travel long distances by foot to see doctors or receive care. In these scenarios AI can fulfill key healthcare worker functions for anyone with an internet connection, in the form of a chatbot — making basic health care in the form of AI immediately available to those who live far from health facilities and medical personnel.

  • AI can bring together data in new ways. By using surveys, satellite images, and health records, we can better predict where people do not have access to vaccines.
  • We can use these insights to make sure vaccines are being delivered when and where they are most needed.
  • AI is increasingly being used to identify better medicines and vaccines faster; the FDA for example saw over 100 submissions using AI in 2021 alone.
  • AI is also being leveraged in vaccine research, particularly in modifying existing vaccines to trigger a stronger immune response with better shelf life — essential for getting shots in arms in environments with challenging supply chains.

Here are seven additional examples of how AI will change healthcare in low- and middle-income countries:

  • AI and machine learning (ML) techniques can analyze data on antimicrobial resistance to identify trends missed by traditional statistical methods.
  • By detecting emerging resistant pathogens and tracking their spread, AI helps public health officials intervene in a timely manner, support research, and create early warning systems.

2. AI in Vaccine Research and Drug Discovery:

  • AI-powered language models provide healthcare workers in low-resource settings with real-time information and advice to inform and support their clinical judgments.
  • These models will use vast amounts of medical knowledge and patient data to improve patient outcomes even in resource-limited environments. Broadly speaking, this could happen either directly or indirectly.
  • Direct — generative AI chatbots can provide accurate, timely, and very cheap medical advice to clients directly to a phone or computer. There are already emerging examples in cancer support, mental health, and acute illness management
  • Indirect — AI can also support and augment healthcare workers in their day-to-day activities. The lowest hanging fruit includes administrative and management tasks such as documentation, medical scribing, and scheduling. AI will also be used as clinical decision support tools, which will assist medical providers and inform their care delivery to patients.

3. Countering Antimicrobial Resistance (AMR):

  • Speech-to-text models can automatically transcribe clinical notes, dramatically reducing the paperwork burden on overstretched health workers.
  • AI and ML techniques can then be applied to those digital notes to classify diseases more accurately, ensure that patients are taking their medication, and create personalized treatment regimes.

4. Empowering Health Workers with Language Models:

  • AI can help estimate local population characteristics and health burdens using existing data sources like surveys. This information helps officials allocate resources efficiently and target health interventions.
  • AI can also predict health-related behaviors, enabling targeted interventions for better outcomes. This can include machine learning and prediction on data from wearables which indicates important behaviors like activity, heart rate, blood pressure, and heart rhythms. AI can also assess and predict aspects of adherence to medical recommendations or medications. This would lead to prompting either the patient for the desired behavior, or deploying clinical support to provide assistance to encourage adherence.

5. Digitizing and Analyzing Clinical Notes:

  • Countries often struggle to analyze and use the data they already have. Governments often turn to outside consultants to integrate datasets, build dashboards, and provide other decision-support tools.
  • AI tools are increasingly capable at data analysis and visualization. Instead of having a consultant create a dashboard or integrate two different data sets, government officials can use natural language to ask an AI chatbot to generate their own dashboards.
  • Donors are often reluctant to fund software engineers and support. Increasingly, however, AI can code. Using these tools can make each engineer far more efficient, increasing governments’ own capacity to scale and maintain digital services.

6. Predicting Local Health Burdens and Behaviors:

7. Democratizing Data Analysis and Interpretation:

As with any transformative technology, we have to mitigate AI’s risks. But there are many people alive today whose lives this technology will improve or save. Public health leaders should welcome responsible use of AI as a powerful tool for good.

Originally published at https://medium.com on June 29, 2023.

--

--

Cooper/Smith

We use hard data to increase effectiveness and efficiency of health and development programs worldwide. www.coopersmith.org