By Emily Vargas. Member of Translating Evidence into Action TWG and CEO of EVidence KBPH SAS
According to Abu-Mostafa, Y. S, Machine Learning (ML) is a branch of Artificial Intelligence (AI) where a machine learns from experience. In its simplest form, the learning algorithms take a bundle of data, analyze them to find certain patterns and, once these are identified, use them to make predictions.
Practical implementation examples
In Mexico Skills – Depot has created an app that identifies, in just a few seconds, a patient’s illness from among more than 200 disease options in its system just from a few signs and symptoms that the clinic’s health worker inputs through an electronic device. It does not just diagnose the illness but, drawing from the hundreds of Clinical Practice Guidelines that the team has inputted into Watson – IBM, the app gives the doctor access to standard clinical management guidelines validated by the Ministry of Health of the country where they are based. The most interesting part of this process is that the algorithm that predicts the best possible answer learns and improves its precision and accuracy the more it is used.
In Colombia, we at EVidence KBPH are developing a software based on ML that sharpens diagnosis of necrotizing enterocolities in newborns by drawing from electronic medical records – quite the challenge. Aside from the high costs of using AI platforms, the ethical dilemmas about the management of clinical histories in the cloud has led us to develop new strategies that allow us to anonymize data without losing the pathway back to the source.
This work is possible thanks to the advances that our partner clinic has made in incorporating Electronic Clinical Histories and standardizing the common language of health – semantic standards. Likewise, the financial support of government entitities, like the Administrative Department of Science, Technology and Innovation (Colciencias), and entrepreneur incubators such as Connect Bogotá Región, who leverage technological developments from new businesses, have played a critical role.
In countries like Mexico and Colombia there are two big actors within the world of AI in health: multinational companies like IBM, Google, Amazon and Phillips, among others, and the entrepreneurs who are seeking to respond to gaps in the sector with their own developments, driven by curiosity and windows of opportunity in this space. Unfortunately, most advances do not reach large markets due to access and implementation challenges: the lack of technological infrastructure, emerging debates over policies, and lack of public trust around AI.
In the health sector the gap between acquisition and development of new technologies, the needed regulation in critical areas, such as interoperability, or the implementation of unified electronic clinical histories, is leaving those of who want to contribute to improving decision-making with the best information resources available without much space to act.
Even so, the landscape is encouraging and there are lessons:
AI is showing us that automation of routine processes for data collection and analysis in health offers competitive advantages for individual health organizations as well as national health systems.
AI offers the opportunity to improve quality of care when it supports faithful implementation of standards and guidelines for clinical care already validated by local authorities.
Big changes do not emerge from legislation—they often go against it—but to the extent that they seek common good and rational use of resources, they could end up modifying legislation itself.
Perhaps AI’s advantages are not recognized by decision-makers in the health sector of our countries, which stalls the advance of implementing these new developments due to lack of clarity in policies and regulations. However, we cannot continue fearing the world of data and information that opened in the 1990s and has been growing immeasurably in the last 10 years. It is time to promote dialogues between organizations and sectors that favor the introduction of new regulations that support thorough incorporation of new technologies in the health systems of our countries.