What is DeepMind? DeepMind is an Artificial Intelligence (AI) company founded in 2010 in the U.K. DeepMind was acquired by Google in 2014. The majority of readers know DeepMind from the famous AlphaGo program that beat the world’s best Go player for the first time. However, DeepMind is said to be also currently working on health-care focused projects that can help in detecting and diagnosing diseases in the early stages.
‘DeepMind and Streams’
DeepMind has developed an application called ‘Streams’, which is aimed at solving the problem of “failure to rescue” where the right health-care professionals do not succeed in treating the patient in time. Streams aims to solve this problem, by providing a system which can immediately process test results. In the case that an issue is found, a notification is sent to alert the health-care professional to help, together with further information regarding the patient. Streams does this by using different types of data and test results from existing IT systems in the respective hospital. Streams has witnessed a lot of success, as nurses say it saves them up to ‘two hours a day’, and patients claiming to have been attended to faster.
‘Streams and Sensitive Data’
However, Google/DeepMind has been under a lot of scrutiny due to Streams. Powles & Hodson (2017) imply that sensitive medical information has been mishandled. As mentioned earlier, Streams relies on existing information from patients to make early diagnoses. However, Streams has access to 1.6 million medical records, and this information was delivered by the National Health Service (NHS) in the U.K. through a data-sharing agreement signed in 2015. Within these medical records lie data that is irrelevant to some of Streams’ services, but critical to patients’ privacy. Therefore, some claim that access to unnecessary data should be restricted.
We know that data plays an important role in the development of A.I such as DeepMind (Machine Learning). But to what extent can A.I be implemented in health-care? How can sensitive data regarding patient’s medical records be properly handled and analyzed? Will we see more A.I in healthcare in the future, and what kind?
Sources:
https://www.forbes.com/sites/bernardmarr/2017/08/08/the-amazing-ways-how-google-uses-deep-learning-ai/2/#50020b4d35e4
https://deepmind.com/applied/deepmind-health/about-deepmind-health/
https://www.theverge.com/2017/3/16/14932764/deepmind-google-uk-nhs-health-data-analysis
Powles, J., & Hodson, H. (2017). Google DeepMind and healthcare in an age of algorithms. Health and Technology, 1-17.
Deniz Ozer
What an interesting article you found, and an interesting blog you wrote about A.L. within healthcare Deniz Ozer! I agree with you that there exists a continuous debate around the consideration between privacy and the use of medical data for future diagnoses and treatments. If you ask me, the use of data to improve diagnoses and treatment has a high priority, and it far more important than the privacy issue. In the end the use of A.L.within the healthcare industry is for the greater good. Furthermore I believe that a massive amount of data can be used whilst still ensuring patient’s privacy. If you think about it, the patient’s identity may be of low importance within a big data set, which is essential to improve the system of deep learning.
You’re article and your question with regard to the potential use of A.L. within the healthcare section gave me an incentive to think about the topic and do some research.Recently I have been to Philips, one of the leading companies within healthcare innovation. During my visit I noticed that the use of data is of high importance within the company. Nevertheless, I noticed that apart from privacy issues also government and market structures influence the possibilities to explore the A.L. applications within the healthcare sector. Thanks or making me more interested in this topic. I am planning on doing some research on the use of A.L. within the healthcare industry and posting an article about it next!
As can be read within the article on Forbes’ website Mariya Yao explains that the US falls behind other countries in advancing with A.L. technologies; not solely because of privacy issues but also due to market structures. She explaines that often the use of A.L. technologies within the healthcare industry are explored due to economic incentives, instead of the improving the outcomes for the patients, and effectively the healthcare of the entire world. As she explained the inter market competition in the US causes companies to be reticent towards sharing data, whilst Canada enforces that all medical records be kept in a central depository owned by the government and citizens, which leads to an enormous data set which could once again benefit the greater good! (Yao, 2017)
To conclude I think instead of asking the questions “Will we see more A.I in healthcare in the future, and what kind?”, we should realize that the use of A.L. and big data (data sharing) in the future could definitely benefit the greater good, and improve our society as a whole. A more important question to ask is; How are we going to mediate and stimulate the use of A.L. within the healthcare industry to achieve our common goal of curing people with illnesses faster, better and more precisely? I believe we should overcome
Literature:
Yao, M. (2017). Forbes Welcome. [online] Forbes.com. Available at: https://www.forbes.com/sites/mariyayao/2017/06/01/u-s-falls-behind-china-canada-in-advancing-healthcare-with-a-i/#85c2804206a3 [Accessed 1 Jun. 2017].
Interesting post Deniz! I certainly believe healthcare will greatly prosper from advances in artificial intelligence (A.I.) next decade and will improve the process of diagnosing, treatment and of course in general, the patientcare. And if you consider at how many innovative firms nowadays are interested in A.I. in healthcare – such as Dell, Apple, Hewlett-Packard, Enlitic, Hitachi Data Systems, Luminoso, Alchemy API, Digital Reasoning, Highspot, Lumiata, and Next IT (Mesko., n.d.) – I think we will definitely see more A.I in healthcare in the future. For example, IBM launched ‘Medical Sleeve’ which is an algorithm to build the next cognitive health assistant to assistant decision-making in clinics in cardiology and radiology. And this is, like DeepMind, just one out of many examples.
However, I believe that the key in using sensitive medical data is finding the balance between using data for health research to benefit society and conducting research in ways that protect the individual’s dignity. Like you said, the public holds a strong privacy concern on how their personal health information is handled, especially the use of data not directly relevant to provide care (Caine et al., 2015; Goolsby, 2011). Therefore I believe that privacy-preserving data mining, pseudonymization and statistical disclosure limitation techniques such as perturbation-, aggregation or cryptographic methods (at least for now) provide a good alternative to handle and analyze sensitive patient information (based on Gostin et al., 2009). This way, sensitive data can be used to benefit health research and protect people’s privacy.
However, I’m wondering how long consumer privacy concerns (the fear of what could done with (un)necessary personal sensitive data) remain to exist. Will it be an issue for the next few decades or will privacy concerns reduce as a consequence of people becoming more familiar with A.I in time? And if so, how long do you think that will take?
References:
Caine, K., Kohn, S., Lawrence, C., Hanania, R., Meslin, E. M., & Tierney, W. M. (2015). Designing a patient-centered user interface for access decisions about EHR data: implications from patient interviews. Journal of general internal medicine, 30(1), 7-16.
Goolsby, A. W., (2011). Clinical Data as the Basic Staple of Health Learning: Creating and Protecting a Public Good: Workshop Summary. National Academies Press.
Gostin, L. O., Levit, L. A., & Nass, S. J. (Eds.). (2009). Beyond the HIPAA privacy rule: enhancing privacy, improving health through research. National Academies Press.
Mesko, B. (n.d.). Artificial Intellgince will redesign healthcare. Accessed on: 14-09-217. Retrieved from http://medicalfuturist.com/artificial-intelligence-will-redesign-healthcare/
Hi Deniz,
Your blog post is touching a very interesting topic, as I think that A.I. will play a big role in health care sooner rather than later. While A.I. in health care still sounds a bit futuristic, it is actually already being used in real-life. I would like to elaborate on one example, which is IBM Watson:
In 2013, IBM Watson launched a program for oncology (preventing, diagnosing and treating cancer). By analyzing hundreds of attributes of a person’s electronic health record (including family history, lab reports and test results), Watson can come up with a diagnosis and personalized treatment options including relevant studies to back up its results. It also suggests which treatment plans should not be used.
Watson is now being used in more than 50 hospitals in 13 countries and the results are pretty amazing (http://www-03.ibm.com/press/us/en/pressrelease/52502.wss):
– In one study, ”Watson for Oncology achieved a concordance rate of 96% for lung, 81% for colon and 93% for rectal cancer cases compared to recommendations from the multi-disciplinary tumor board in a study at Manipal Comprehensive Cancer Centre in Bangalore, India.”
– Another study showed that ”Watson for Oncology achieved a concordance rate of 83% for multiple cancer types compared to recommendations from oncologists in a study at Bumrungrad International Hospital, a multispecialty hospital in Bangkok, Thailand.”
– The time it takes to screen patients is also massively reduced by using Watson, as ”Watson for Clinical Trials Matching cut the time required to screen patients for clinical trial eligibility by 78% in a technology feasibility study with Highlands Oncology Group and Novartis”.
According to IBM, Watson for Oncology has already touched 14,000 people around the world and I think Watson for Oncology is just a tip of the iceberg of what A.I. has in store for the health care industry.
If you want to know a little bit more about how Watson for Oncology works, check out this video: https://youtu.be/8_bi-S0XNPI