Your Health and Artificial Intelligence.
How safe would you feel if you were diagnosed and treated by artificial intelligence by a machine, so to speak?
First, let us take a look at the definition of artificial intelligence and what it is all about.
Artificial intelligence (AI) is the ability of a computer program or a machine to think and learn. It is also a field of study which tries to make computers “smart”. They work on their own without being encoded with commands. John McCarthy came up with the name “artificial intelligence” in 1955.
In general use, the term “artificial intelligence” means a machine that mimics human cognition. “The theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages.”
There’s no doubt that technology has changed the way health care happens—everywhere.
“Even in rural Uganda, a patient can confirm her provider’s authenticity with a text message, and a community health worker can use Google to learn symptoms and treatments,” says Wayan Vota, IntraHealth’s director of digital health.
Now artificial intelligence (AI) and machine learning are changing the way we manage our health in and outside the clinic. We’re starting to see more instances where health workers and researchers can use AI to diagnose eye disease, depression, Alzheimer’s disease, and more.
And then there’s the DIY health tech. Want someone to talk to? There’s a chatbot therapist app for that. Want to have AI on your computer analyze your keystrokes and predict whether you’re developing a neurodegenerative disorder? You can sign up for that here. Want to track and record your every move to stay fit? Keep reading. All these new tools and applications are changing the way we take care of ourselves. They’re also generating scads of health data, which present their own challenges.
What is Artificial Intelligence in Healthcare?
Machine learning has the potential to provide data-driven clinical decision support (CDS) to physicians and hospital staff – paving the way for increased revenue potential. Machine learning, a subset of AI designed to identify patterns, uses algorithms and data to give automated insights to healthcare providers.
Examples of AI in Healthcare and Medicine
AI can improve healthcare by fostering preventative medicine and new drug discovery. Two examples of how AI is impacting healthcare include:
IBM Watson’s ability to pinpoint treatments for cancer patients, and Google Cloud’s Healthcare app that makes it easier for health organizations to collect store, and access data.
Business Insider Intelligence reported that researchers at the University of North Carolina Lineberger Comprehensive Cancer Center used IBM Watson’s Genomic product to identify specific treatments for over 1,000 patients. The product performed big data analysis to determine treatment options for people with tumors who were showing genetic abnormalities.
Comparatively, Google’s Cloud Healthcare application programming interface (API) includes CDS offerings and other AI solutions that help doctors make more informed clinical decisions regarding patients. AI used in Google Cloud takes data from users’ electronic health records through machine learning –creating insights for healthcare providers to make better clinical decisions.
Google worked with the University of California, Stanford University, and the University of Chicago to generate an AI system that predicts the outcomes of hospital visits. This acts as a way to prevent readmission and shorten the number of times patients are kept in hospitals.
Benefits, Problems, Risks & Ethics of AI in Healthcare
Integrating AI into the healthcare ecosystem allows for a multitude of benefits, including automating tasks and analyzing big patient data sets to deliver better healthcare faster, and at a lower cost. According to Business Insider Intelligence, 30% of healthcare costs are associated with administrative tasks. AI can automate some of these tasks, like pre-authorizing insurance, following-up on unpaid bills, and maintaining records, to ease the workload of healthcare professionals and ultimately save them money.
AI has the ability to analyze big data sets – pulling together patient insights and leading to predictive analysis. Quickly obtaining patient insights helps the healthcare ecosystem discover key areas of patient care that require improvement. Wearable healthcare technology also uses AI to better serve patients. Software that uses AI, like FitBits and smartwatches, can analyze data to alert users and their healthcare professionals on potential health issues and risks. Being able to assess one’s own health through technology eases the workload of professionals and prevents unnecessary hospital visits or remissions.
As with all things AI, these healthcare technology advancements are based on data humans provide –meaning, there is a risk of data sets containing unconscious bias. Previous experiences have shown that there is potential for coder bias and bias in machine learning to affect AI findings. In the sensitive healthcare market, especially, it will be critical to establish new ethics rules to address – and prevent – bias around AI.
Future of Artificial Intelligence in Healthcare
The use of AI in the healthcare market is growing due to the continued demand for wearable technology, digital medicine, and the industry’s overall transformation into the modern, digital age. Hospitals and healthcare professionals are seeing the benefits of using AI in technology and storing patients’ data on private clouds, like the Google Cloud Platform. AI allows doctors and patients to more easily access health records and assess patient’s health data that is recorded over a period via AI-infused technology
Health tech companies, startups, and healthcare professionals are discovering new ways to incorporate AI into the healthcare market; and, the speed at which we
improve the healthcare system through AI will only continue to accelerate as the industry dives deeper into digital health. Artificial intelligence in health care carries huge potential, according to experts in computer science and medicine, but it also raises serious questions around bias, accountability, and security.
“I think we’re just seeing the tip of the iceberg right now,” said Yoshua Bengio, a computer scientist and professor at the University of Montreal, who was recently awarded the Turing Award, often called the “Nobel Prize” of computing. Bengio is one of the pioneers of deep learning, an advanced form of AI, which he believes will advance health care. In deep learning, a computer is fed data, which it uses to make assumptions and learn as it goes — much like our brain does.
Scientists are already using AI to develop medical devices. At the University of Alberta, researchers are testing an experimental bionic arm that can “learn” and anticipate the movements of an amputee. Last year, the U.S. Food and Drug Administration (FDA) approved a tool that can look at your retina and automatically detect signs of diabetic blindness.
Emergency Room Waiting Times
At Humber River Hospital in northwest Toronto, AI is speeding up perhaps the most frustrating part of a patient’s experience: the emergency room. In the hospital’s control center, powerful computers are now accurately predicting how many patients will arrive in the emergency department — two days in advance.
The software processes real-time data from all over the hospital— admissions, wait times, transfers, and discharges — and analyzes it, going back over a year’s worth of information. From that, it can find patterns and pinpoint bottlenecks in the system. “If you add up all those tiny delays — how long it takes to see your doctor, how long you’re waiting for your bed to be cleaned, how long you’re waiting to get up to your room — if you measure all of those things and can shorten each one of them, you can start saving a lot of money,” said Dr. Michael Gardam, chief of staff at Humber River Hospital.
According to Gardam, it’s working: patients are now moving through the system faster, allowing the hospital to see an average of 29 more patients a day.
Risks With AI and Health
For machines to learn, they need vast amounts of information. Since that initial data comes from humans, some of that information can be tainted by personal bias especially if the algorithm isn’t fed a diverse data set.
“In dermatology, you take a look at a number of different photographs or slides of moles. If you happen to be pale-skinned, some of the machine learning associated with that imagery is great. If you’re darker-skinned, it’s not,” said Dr. Jennifer Gibson, a bioethicist at the University of Toronto. She’s not against the integration of AI in health care but warns that anything involving big data, profit-driven companies, and health care should be heavily regulated.
“In our hunger for more data, in order to power these tools, we may be introducing a form of surveillance within our society — which is not really the intended goal,
but might happen accidentally,” Gibson said.
Gardam doesn’t share those concerns; he believes humans — not machines — will remain in control.”It’ll still be a long time before we fully accept information coming from a computer system, telling us what the diagnosis is,” he said. “Humans are still going to be reviewing it until we’re very comfortable we’re not
Some governments aren’t waiting for that to happen. In the U. S, the FDA recently announced that it is developing a framework for regulating self-learning AI products used in medicine. In a statement to CBC News, Health Canada said it also engaging with national, international, industry, academic, and government stakeholders “to discuss the challenges and opportunities in regulating current and emerging AI technologies in health care.”
What The 21st Century is Bringing To Us As Far Health Care and AI
In the 21st Century, the age of big data and artificial intelligence (AI), each healthcare organization has built its own data infrastructure to support its own needs, typically involving on-premises computing and storage. Data is balkanized along organizational boundaries, severely constraining the ability to provide services to patients across a care continuum within one organization or across organizations.
This situation evolved as individual organizations had to buy and maintain the costly hardware and software required for healthcare, and has been reinforced by vendor lock-in, most notably in electronic medical records (EMRs). With increasing cost pressure and policy imperatives to manage patients across and between care episodes, the need to aggregate data across and between departments within a healthcare organization and across disparate organizations has become apparent not only to realize the promise of AI but also to improve the efficiency of existing data-intensive tasks such as any population-level segmentation and patient safety monitoring.
The rapid explosion in AI has introduced the possibility of using aggregated healthcare data to produce powerful models that can automate diagnosis and also enable an increased approach to medicine by tailoring treatments and targeting resources with maximum effectiveness in a timely and dynamic manner.
However, “the inconvenient truth” is that at present the algorithms that feature prominently in the research literature are in fact not, for the most part, executable at the front lines of clinical practice. This is for two reasons: first, these AI innovations by themselves do not re-engineer the incentives that support existing ways of working.
A complex web of ingrained political and economic factors and the proximal influence of medical practice norms and commercial interests determine the way healthcare is delivered. Simply adding AI applications to a fragmented system will not create sustainable change. Second, most healthcare organizations lack the data infrastructure required to collect the data needed to optimally train algorithms to (a) “fit” the local population and/or the local practice patterns, a requirement prior to deployment that is rarely highlighted by current AI publications, and (b) interrogate them for bias to guarantee that the algorithms perform consistently across patient cohorts, especially those who may not have been adequately represented in the training cohort.
For example, an algorithm trained on mostly Caucasian patients is not expected to have the same accuracy when applied to minorities. In addition, such rigorous evaluation and re-calibration must continue after implementation to track and capture those patient demographics and practice patterns which inevitably change over time.
Some of these issues can be addressed through external validation, the importance of which is not unique to AI, and it is time that existing standards for prediction model reporting are being updated specifically to incorporate standards applicable to this end. In the United States, there are islands of aggregated healthcare data in the ICU, and in the Veterans Administration. These aggregated data sets have predictably catalyzed an acceleration in AI development, but without broader development of data infrastructure outside these islands, it will not be possible to generalize these innovations.
The Google Cloud, Health and AI
Elsewhere in the economy, the development of cloud computing, secure high-performance general use data infrastructure and services available via the Internet (the “cloud”), has been a significant enabler for large and small technology companies alike, providing significantly lower fixed costs and higher performance and supporting the aforementioned opportunities for AI. Healthcare, with its abundance of data, is in theory well-poised to benefit from growth in cloud computing. The largest and arguably most valuable store of data in healthcare rests in EMRs. However, clinician satisfaction with EMRs remains low, resulting in variable completeness and quality of data entry, and interoperability between different providers remains elusive.
The typical lament of a harried clinician is still “why does my EMR still suck and why don’t all these systems just talk to each other?” Policy imperatives have attempted to address these dilemmas, however, progress has been minimal. In spite of the widely touted benefits of “data liberation”, a sufficiently compelling use case has not been presented to overcome the vested interests maintaining the status quo and justify the significant upfront investment necessary to build data infrastructure.
Furthermore, it is reasonable to suggest that such high-performance computing work has been and continues to be beyond the core competencies of either healthcare organizations or governments and as such, policies have been formulated, but rarely, if ever, successfully implemented. It is now time to revisit these policy imperatives in light of the availability of secure, scalable data infrastructure available through cloud computing that makes the vision of interoperability realizable, at least in theory.
To realize this vision and to realize the potential of AI across health systems, more fundamental issues have to be addressed: who owns health data, who is responsible for it, and who can use it? Cloud computing alone will not answer these questions—public discourse and policy intervention will be needed. The specific path forward will depend on the degree of a social compact around healthcare itself as a public good, the tolerance to a public-private partnership, and crucially, the
public’s trust in both governments and the private sector to treat their healthcare data with due care and attention in the face of both commercial and political perverse incentives.
In terms of the private sector, these concerns are amplified as cloud computing is provided by a few large technology companies who have both significant market power and strong commercial interests outside of healthcare for which healthcare data might potentially be beneficial. Specific contracting instruments are needed to ensure that data sharing involves both necessary protection and, where relevant, fair material returns to healthcare organizations and the patients they serve. In the absence of a general approach to contracting, high profile cases in this area have been corrosive to public trust.
Data privacy regulations like the European Union’s General Data Protection Regulation (GDPR) or California’s Consumer Privacy Act are necessary and well-intentioned, though incur the risk of favoring well-resourced incumbents who are more able to meet the cost of regulatory compliance thereby possibly limiting the growth of smaller healthcare provider and technology organizations.
Initiatives to give patients access to their healthcare data, including new proposals from the Center for Medicare and Medicaid Services are welcome, and in fact, it has long been argued that patients themselves should be the owners and guardians of their health data and subsequently consent to their data being used to develop AI solutions.
In this scenario, as in the current scenario where healthcare organizations are the de-facto owners and guardians of patient data generated in the health system alongside fledgling initiatives from prominent technology companies to share patient-generated data back into the health system, there exists the need for secure, high-performance data infrastructure to make use of this data for AI applications.
If the aforementioned issues are addressed, there are two possible routes to building the necessary data infrastructure to enable today’s clinical care and population health management and tomorrow’s AI-enabled workflows. The first is an evolutionary path to creating generalized data infrastructure by building on existing impactful successes in the research domain such as the recent Science and Technology Research Infrastructure for Discovery, Experimentation, and Sustainability
(STRIDES) an initiative from the National Institutes of Health or MIMIC from the MIT Laboratory for Computational Physiology to generate the momentum for change.
Another, the more revolutionary path would be for governments to mandate that all healthcare organizations store their clinical data in commercially available clouds. In either scenario, existing initiatives such as the Observational Medical Outcomes Partnership (OMOP) and Fast Healthcare Interoperability Resources (FHIR) standard that create a common data schema for storage and transfer of healthcare data and AI-enabled technology innovations to accelerate the migration of existing data will accelerate progress and ensure that legacy data are included.
There are several complex problems still to be solved including how to enable informed consent for data sharing, and how to protect confidentiality yet maintain data fidelity. However, the prevalent scenario for data infrastructure development will depend more on the socioeconomic context of the health system in question rather than on technology.
A notable by-product of a move of clinical and research data to the cloud would be the erosion of the market power of EMR providers. The status quo with proprietary data formats and local hosting of EMR databases favors incumbents who have strong financial incentives to maintain the status quo. The creation of health data infrastructure opens the door for innovation and competition within the private sector to fulfill the public aim of interoperable health data.
The potential of AI is well described, however in reality health systems are faced with a choice: to significantly downgrade the enthusiasm regarding the potential of AI in everyday clinical practice, or to resolve issues of data ownership and trust and invest in the data infrastructure to realize it.
Now that the growth of cloud computing in the broader economy has bridged the computing gap, the opportunity exists to both transform population health and realize the potential of AI, if governments are willing to foster a productive resolution to issues of ownership of healthcare data through a process that necessarily transcends election cycles and overcomes or co-opts the vested interests that maintain the status quo—a tall order. Without this, however opportunities for AI in healthcare will remain just that—opportunities.
Panch, T., Szolovits, P. & Atun, R. Artificial intelligence, machine learning, and health systems. J. Glob. Health 8, 020303 (2018).
A., Michalowski, M. & Buckeridge, D. Health intelligence: how
artificial intelligence transforms the population and personalized health. NPJ Digit Med. 1, 53 (2018).
Fogel, A. L. & Kvedar, J. C. Artificial intelligence powers digital medicine. NPJ Digit Med. 1, 5 (2018).
I wonder whether to be worried about this or accept it with confidence. Whatever goes into artificial intelligence is provided by humans. How would this work with medical bills? What about the real Doctor?
I think about the incorporation of AI into automobiles. Again Google is right there amongst other AI automobiles like Tesla.
The autopilot system has been around for several years, but its functionality is relevant to flying the plane once it is already in the sky and everything is going smoothly. It can not land the plane or deal with complications if there were problems.
Will this be like the health care system of the future?
Thank you for reading,
Comments are welcome