Big data and especially Artificial intelligence will continue to transform health care through early disease detection as well as anomaly pattern identification, prognosis and, reaction time enhancements. A study conducted by Nyu researchers reveal how incorporating online data in your studies like using google street view images can be useful in public health research but at the same time points out its shortcoming if used individually.
From the article which sourced data from the Proceedings of the National Academy of Sciences, the writer concentrated on two million images by the Google Street View in order to ascertain how big data assists in the enhancement of public health.
The new data sources help in increasing the level of comprehensiveness in explaining health since it includes additional features like applying machine learning and data science to derive the insights from the data. ” Rumi Chunara, the senior author and an associate professor at New York University. But apparently, Chunara pointed out that, relying on such digital sources may result in a number of challenges.
Therefore, the developed conclusions show how the connections between environmental impacts, behavior, and health effect may be complex and can be more than can be described within the restrictive framework of AI ideas.
Chunara and the research team including a Ph. D. student Miao Zhang used AI to develop the indicator of sidewalk and crosswalk accessibility based on street view imagery. For comparison with this data, local data on obesity, diabetes and physical activity collected from the CDC were also considered.
While this brought about a positive correlation in the probability of cross walks reducing the mean obesity and diabetes index in the community, availability as determined by sidewalks did not have a correlation with health statuses, unlike other studies.
But the author stressed that many of sidewalks are in areas where people rarely walk like highways or bridges and could explain why sidewalks provision seems to have little correlation with health indicators. The researchers also indicated some of the potential issues with the AI assigned labels for street view images; often these do not suggest the presence of sidewalks due to cars or the shadow.
The research conducted by the researcher concluded that it was the exercise that was a necessity and not the cross walks because these are the key areas that helped in preventing or at least minimizing the cases of obesity and diabetes. By increasing physical activity, clearly the benefits seen were much better than having more crosswalks put up across the country.
According to the research, this can be attributed to the fact that the researchers recommend increasing encouragement of physical activities such as aerobics that are community-based than relying on the modification of structures.
Their findings raise the need to use big data together with domain expertise in public health and computer science. New data sources don’t allow relying on associations only to get the best interventions or to use resources most effectively. The integration of AI with domain knowledge might be the better solution to be used to influence the concept and recommend the correct decision that will create an improved public health.
Reference
Miao Zhang et al, Utilizing big data without domain knowledge impacts public health decision-making, Proceedings of the National Academy of Sciences (2024). DOI: 10.1073/pnas.2402387121.
