The use of big data and increasingly accurate analytical tools in the medical field could help improve patient care and treatment, reduce costs associated with health services, predict and prevent epidemic phenomena. Big data have changed and will change even more in the future many areas of health and medicine: from diagnostics to care to the management of health services. The insights gained from this huge amount of information, in health care can help researchers to gain a deeper understanding of the phenomena in order to improve the results of clinical trials, increase the productivity of health workers and improve the effectiveness of the practices themselves. On these topics we interviewed Russ B. Altman, professor of bioengineering, genetics, medicine, and biomedical data science (and of computer science, by courtesy) and past chairman of the bioengineering department at Stanford University.
In your opinion, what were the main advances made by big data in the healthcare sector?
We can see the emergence of three big data sources: genomics, electronic medical records,wearables, sensors and mobile health. These have combined to give a huge opportunity to make measurements about patients and use it for their health. These combine with the second advance, Artificial Intellicence (A.I.) and machine learning, which can extract knowledge from these data.
Are there limits to the current use of big data in healthcare?
Noisy data which obscures signal and biased data which gives wrong signals are both big problems. In addition, we are sometimes making measurements that we don’t have good knowledge or evidence about how to interpret, so our measurements are outpacing our ability to interpret them.
How will it be possible to balance personalized medicine, large amounts of patient data and privacy?
The whole point of personalized medicine is to base it on detailed measurements of the patient. So we need to develop social and technical means to protect the data. On the social hand, it means laws and regulations that specify penalties for violation of privacy. On technical hand, we need methods for making it difficult to break into and use private data. There, we expect homomorphic encryption, blockchain and similar technologies to be very helpful.
How should bigdata be used with modeling and artificial intelligence to make the healthcare sector even more futuristic?
The future is now. Eric Topol wrote a great review of all the ways in which AI is starting to impact medicine. It will happen and as long as we remember the welfare of the patient (and not only costs and finances), we should be able to improve healthcare and health outcomes. Keep people healthy longer and treat disease more effectively with appropriate AI technologies. Also, it will be important to use AI to make sure that healthcare remains compassionate humans when necessary and AI when best.
What research areas need to be improved to enable bigdata to contribute even more to healthcare?
Stanford has a new Institute for AI called human-centered artificial intelligence and it outlines three good areas for improved research: human-centered research on societal, human, cultural, economic impacts of AI in health; articial intelligence algorithms that are based on huma reasoning and hopefully therefore interact better with human thinkers; applications that augment human capability and do not replace humans, or if they do replace humans, do it in a way that creates more human opportunity for work and play.
Are there any misunderstandings about the relationship between bigdata and healthcare? If yes, which ones?
The main misunderstanding is that although we have much more data than previously, it is still nothing compared to the data that big internet companies like Google, Facebook, Twitter, Netflix have. So we still are “data poor” for many applications and must remember that brute force methods of using data will not work, we still need to interpret and use our theoretical knowledge to put data in proper perspective.