More than 80% of the world’s population currently lives in low- and middle-income countries. Regrettably, the current model of the pharmaceutical industry offers weak incentives for companies to engage in these markets, which explains why 2 billion people still lack access to basic medicines across the globe.
The advent of artificial intelligence as an adaptive and predictive technology offers the possibility for radical optimization of business practices, reshaping market opportunities for pharmaceutical companies and ultimately challenging the status quo on access to medicine worldwide.
However, pharmaceutical companies are only starting to explore AI fields of application. As reported by Microsoft, more than 50% of AI professionals work for the tech sector, while only 3% currently work for health organizations and 1% for nonprofit organizations. This disproportion certainly hampers the speed and extent of the technology’s adoption within the pharmaceutical sector.
Furthermore, the first exploratory usages are concentrated in areas where costs and timelines can be minimized — namely in drug discovery and development. Even though it is standard behavior for a corporation to adopt such a strategy, pharmaceutical companies can and should expand their horizons by using these new technologies in creating value for LMICs. Cost savings and increased automation can leapfrog the main issues of access, such as the lack of scalability and low profit margins.
COVID-19 is a litmus test for whether AI can help in the crisis response. So far, the technology has already played a role in many aspects of the pharmaceutical industry. It has helped scientists access precise information by analyzing thousands of scientific papers on the topic, it supported the prediction of old and new drugs and treatments that might alleviate the disease, it enabled the rapid diagnosis of patients with COVID-19, and more. As the pandemic intensifies, can AI also support equitable access of the vaccine in LMICs?
The process is simple: Record your cough and answer a short survey detailing your current symptoms, location, gender, and other diseases. You could be helping detection efforts.
There are multiple AI-driven interventions in which the pharmaceutical industry can have a significant influence. For example, current inefficiencies and weaknesses along supply chains — whether in procurement processes, delivery logistics, traceability, or storage — not only impact the accessibility, availability and quality of medicines but also represent significant costs to companies while exposing them to reputational issues.
This is particularly important as many LMICs have a limited local manufacturing capacity of essential medicines. The whole continent of Africa, for example, relies on imports for 80% or more of its pharmaceuticals and medical consumables.
Three main routes for AI activities related to the supply chain and access to medicine have been identified:
1. Demand forecasting
The demand for specialized pharmaceuticals in most countries of sub-Saharan Africa is low and unpredictable, necessitating excessive inventory holdings to secure in-country availability. This, in turn, increases storage cost and product obsolescence. Using AI-based demand-forecasting tools helps anticipate shifts in demand and consumption, decreasing stockouts and wastage while ultimately increasing opportunities for immunization.
Global coverage of the basic childhood vaccine for diphtheria, tetanus, and pertussis, for example, has reached 85%. Yet, vaccine wastage remains an important issue. One report found that for every 100 vaccine vials used in Tanzania, supply chains deliver between 43 too many and 43 too few vials. On the one hand, this puts lives at risk, while on the other hand, unnecessary wastage occurs.
Innovations from previous crises provide lessons for our near-term response, but also highlight the underlying and more critical long-term need to better resource and institutionalize strategic innovation in global health. This op-ed explains how this can be done.
Demand is not a single-dimensional issue, and it can be impacted by a range of factors. By using data points from multiple sources — the majority of which are not health-related and include socioeconomic and demographic considerations — AI company Macro-Eyes’ machine-learning technology was able to reduce the misallocation to 2.42 vials per 100 in Tanzania, without relying on the country’s registry or facility utilization data.
2. Reporting substandard medicines
Substandard medicines cause harm to people, as well as death. The World Health Organization estimates that 1 in 10 medicines in LMICs are estimated to be substandard or falsified, which is why it has established the Medical Product Rapid Alert. Reporting confirmed cases of substandard medicines in a timely manner to the relevant authority is crucial to allow withdrawal from the market as quickly as possible.
To protect citizens, regulators impose testing for imported pharmaceutical products, which unfortunately delays supply to needy patients and increases supply chain costs. By using AI techniques, a specialist may set statistical limits to reduce the testing samples and decrease the testing period while conforming to regulatory requirements.
Furthermore, should a defect arise, the use of AI on traceability and serialization could allow pharmaceutical companies to pinpoint exactly where any contamination or defect originated in the supply chain, allowing teams to correct around the issue more efficiently without ordering massive recalls of drug batches, which would create cascading roadblocks to patient care. AI, combined with analytics, helps make precise decisions while improving efficiencies.
3. Ensuring continuous supply
Ensuring continuous supply and providing emergency humanitarian relief are often compromised by the complexity of supply chains. AI systems have the ability to provide autonomous supply operation where decisions about allocating materials and distributing products are being made, from active pharmaceutical ingredient shortages to temperature management.
Winners of the Google AI Impact Challenge Summit say that while AI can help address societal problems, humans need to be involved, rather than having computers make decisions for people.
Sensors and algorithms matched throughout the supply chain allow for real-time visibility of supply chain performance while accounting for reserves. In the event of a natural catastrophe destroying a drug supply, for example, the system could determine the most efficient way to produce or distribute the product, as needed.
These systems would allow the prevention and mediation of global drug shortages, such as a shortage of benzathine penicillin that affected more than 39 countries in 2017, putting millions of lives at risk across the globe.
Developing autonomous and intelligent supply operations adapted to the realities of low- and middle-income countries is a hard task. However, “94% of healthcare executives report that the pace of innovation in their organization has accelerated over the past three years due to emerging technologies,” according to the “Digital Health Tech Vision” report.
These include so-called extended reality, quantum computing, distributed ledger technology, and AI. Of all four, 41% of health care executives named AI as the one technology with the greatest impact on their organization over the next three years. Its ability to automate business functions and scan unprecedented amounts of data offers promising potential.
There is no one-size-fits-all strategy for improving access to medicine in LMICs. As the pharmaceutical industry furthers its exploration in AI, companies investing in and leveraging the power of this technology for access could be at the forefront of what may set a new standard for the industry.