AI is helping treat healthcare as if it’s a supply chain problem

To support MIT Technology Review journalism, please consider becoming a subscriber.,

Monitors whether the recommendations of the FIND tool are useful as well as whether they are put into practice. “This needs to be something that countries can take ownership of,” says Albert. “One of the main things is to put the power of data in the hands of the people [who] Decisions need to be made. “

Catch up to the smartphone

BroadReach, in partnership with Microsoft, uses a simulation tool called Vantage, which identifies non-staffed clinics and sends health care workers where they are most needed. In 2020, during the first few weeks of the epidemic, BroadReach worked with FIND on how much the two South African provinces were ready for Kovid, demonstrating the lack of protective equipment and staff in more than 300 clinics in just three days.

The sergeant says he first learned about health care systems in Africa while working in refugee camps before going to medical school. He later co-founded Broadrich with fellow doctor Ernest Darkoh, who grew up in Tanzania and Kenya. “You can go to rural clinics in places like Zambia, and you’ll see patient records sitting on paper,” says the sergeant. The technology is there though: “Nurses are using smartphones and Facebook is suggesting posts they like.”

In addition to monitoring shortages, Broadrich tracks individual patients in more than a thousand clinics in some African countries, monitors whether they are receiving the treatment they need, and continues to comply. Clinics already do that, but also use Vantage Machine Learning, trained on thousands of anonymous clinical records and social data, to suggest that people are likely to drop out of treatment and that health care workers actively check with them.

The Institute of Virology Nigeria used Ventage in 2021 to predict which of the 30,000 people receiving treatment for HIV at three locations in Nigeria were at risk of discontinuation of their drug. The tool found that as a result, 91% of those who received a call or visit from a health care worker were up-to-date with their medication, compared to 55% who were not contacted.

According to Broadrich, health care workers at a number of HIV clinics say the tool helps them maintain close relationships with their patients and help them focus on those who need intervention the most.

The so-called developed world

Broadrich now wants to make its software available in the US. “Around the time of Kovid’s hit, we woke up to the fact that a lot of the Quote-Unquote developed-developed-world health systems weren’t so good, and most of the population was left behind,” says Sergeant.

Broadrich is involved in four pilot projects with US healthcare providers and insurance companies. In one, he encountered low rates of vaccination in parts of Colorado using machine learning, where vaccination sites should be set up and what communities should be promoted. Local health officials assumed that resources should be concentrated in urban areas to vaccinate as many people as possible. But Ventage revealed that focusing on low-income, rural, minority communities would have a greater impact.

Broadrich is also working with an insurer in California who sees significant disparities in the way people from different groups treat statins, the drugs used to treat high cholesterol. Looking at the data, Broadrich seeks to identify possible explanations for what the insurer is looking for. Some communities have poor transport links to clinics, which can prevent people from visiting their doctor to update prescriptions. The sergeant says others have long had distrust of the health care system.

Ultimately he wants to see if Vantage predicts risk factors for individual patients. For example, for a Spanish speaker who does not live near the clinic, the software would recommend that the insurer provide a voucher for a Spanish-speaking social worker and a taxi, he says.

But it is difficult to capture the data needed to train AI to make such predictions accurate. In the US, health care data is not generally shared between providers. The sergeant says Broadrich is getting around this by linking medical records to socio-economic data, such as people’s zip codes and credit histories. “We have partnerships with consumer data companies, because if you have a pattern of patient behavior and the conditions in which they live, you can say a lot about it,” he says. “We combine all of this to see the overall population and each patient’s view.”

How people feel about this type of supervision depends on what really benefits them. Broadrich already uses socio-economic data to predict the potential future behavior of individuals – a range of organizations, including lending companies, rental agencies, police and more. Prejudices in these systems have led to fairly strong pushbacks from civil rights groups.

Nicholson Price, who is studying legal and ethical questions about the use of personal data at the University of Michigan, says government proposals to share medical data have provoked reactions in some countries, including the US, UK and Australia. But that doesn’t stop companies from linking medical and consumer data. “Companies have been doing this for years, just on a low profile,” says Price.

“There’s a sense of resignation that it’s happening and we don’t have the capacity to stop it,” he says. “That said, maybe there’s a silver lining that some good stuff will come out of this too, instead of just being advertised and tampered with.”

Similar Posts

Leave a Reply

Your email address will not be published.