Sentry Page Protection
Medicine Interview Hot Topics: Digitisation of Healthcare Data & AI in Healthcare
As healthcare advances, previous staples of the doctor’s working life become part of their memory. Gone is the notebook - replaced with electronic medical records (EMR). As medical data is written and stored, it becomes increasingly pertinent to ask, who should have that data? Should it be combined to aid learning and AI? Is it better in the hands of local healthcare providers or as a public resource?
The Current Situation
As it stands, each individual healthcare organisation has its own data infrastructure, which supports its own needs - with little thought given to the bigger picture. Data is stored in-organisation, using rules and formats for that organisation. The ability to maintain care across organisations is severely constrained. Individual organisations will typically have bought their own EMR systems, which are locked by the provider of that system. Whilst countries with unified healthcare are likely to have a single system that all doctors are used to, no matter the practice, the US therefore has numerous different systems. This doesn’t fit with the increasing push to provide cheaper, more efficient healthcare, and the desire to allow a true continuity of care between hospitals, practices and areas. It seems that aggregating data between different areas of healthcare will benefit each organisation, individual patients, and allow the use of AI.
What Should Change?
As the capability of AI has exploded in recent years, it becomes more possible to use aggregated healthcare data to train models that will automate diagnosis, target resources and recommend treatments, using a depth of knowledge that would otherwise be impossible to achieve. This would be a great benefit as an adjunct to the doctor’s clinical judgement. However, as well as the correct AI systems, we would need the correct data infrastructure to allow this process to begin. It would require the sharing of information between practices, an understanding of the local population, and a guarantee that the algorithm will perform to the same level through different cohorts - meaning that if the patient data lacks one type of patient on which to train the AI, this must be known and understood. Likewise, re-calibration would have to continue after the AI had already been implemented.
As it stands, the average clinician is unsatisfied with their EMR. They struggle to enter basic data, finding that the system makes simple tasks complicated. There is no communication between one system and another. It therefore is apparent that we cannot expect the average doctor to wholeheartedly embrace the chance to use AI and aggregated data in their practice.
However, we can encourage them to consider who owns healthcare data, and who should have the right to use it. The public needs to trust that the government will use their data for the right reasons, and doctors having an interest in, and understanding of, data protection and the use of data is crucial if people are to believe that their healthcare data is being used for their best interests.
If the US government mandated that each healthcare had to store their clinical data in a secure cloud, would this be of benefit? We must question the huge security risks associated with such a move, and balance them against the potential upside of the ability to aggregate and study this data in a manner never achievable before.
If we are to realise the potential of AI to revolutionise healthcare, we must resolve the issue of who owns data, and the government must invest in data infrastructure. With more investment in data infrastructure, and side-by-side investment in EMR systems, doctors could have access to a patient system that not only functions well for them, but provides massive benefit to them, or their colleagues, when conducting research.
Example Interview Questions
- What do you think about digitisation of healthcare in the US?
- What are some of the issues around digitisation of healthcare in the US?
- Who owns healthcare data in the US?
- What is an EMR?
- What would the benefits of aggregating healthcare data in the US be?
- What are the risks of aggregating healthcare data?
- How might you convince a traditional family doctor to subscribe to a new EMR system that will aggregate his data with that of many others?
- In the future, how would you describe data aggregation and AI to a patient curious about what happens to their healthcare data?
- What do you think the impact of AI will be on healthcare?
How to answer questions on AI in healthcare
Remember that you don’t need to know a great deal about this topic - having a basic understanding is sufficient - but you can stand out through knowing about it. If you are considering talking about public health or research you should especially consider the roles of digitisation, aggregation of data, and the use of AI.
Interview Questions & Example Answers
What do you think about digitisation of healthcare in the US?
I believe that the digitisation of healthcare is crucial to it making progress. As it stands individual organisations are storing their data in their own ways, and in many cases hospitals still rely on outdated technologies like faxes to transfer information, such as scans or results. I therefore consider that a universal push for digitisation, and the use of the same types of systems nationwide, is vital. If we continue down this route, with EMR systems chosen at a national level, then data could be accessed by AI systems and used at scale to promote healing and conduct research.
What are some of the issues around digitisation of healthcare in the US?
As it stands, the data infrastructure to share information between different family practices and general hospitals is not in place. In order to digitise information at scale and use it for research, information must be available on all types of patients, in order to avoid biases. Clinicians currently are not satisfied with their EMR systems, as they find them difficult to use and needlessly complex - making simple tasks take far too long. Individual EMR systems have no method of communication with other systems. Additionally, issues over who owns healthcare data are complex and must be fully resolved before it can be used on a national scale.
Who owns healthcare data in the US?
Under both federal and state law, patients have legal rights related to their healthcare data. However, when their data is captured and documented - be it written on paper or electronically - the healthcare provider gains the right of possession of that data, as they own the media on which it is stored. Therefore, the healthcare provider that the patient uses is technically the owner of their data. The patient, however, retains the right to access their data and request that changes or copies be made.
What is an EMR?
An EMR is an electronic medical record - a digital version of the paper records that previously would have been stored in a doctor’s office. It contains the medical and treatment history of the patient in that practice specifically. Advantages of EMRs over a traditional paper-based system include that they can track data over time, allow for the identification of patients that are overdue screenings or check-ups, check and filter by parameters like blood pressure, and provide an overall method of monitoring the quality of care being provided by the practice.
What would the benefits of aggregating healthcare data in the US be?
Aggregated healthcare data would allow firstly for more efficient patient care and improved outcomes as practices, hospitals and tertiary providers of care could more readily access the patient information that they required. Additionally, it would allow for the training of models that automatically diagnose, recommend treatments or target resources. This would act as an adjunct to doctor’s clinical judgement, drawing upon a quantity of data that has never been previously accessible for machine learning or any other form of study. New research could be performed, again at a scale never before seen.
What are the risks of aggregating healthcare data?
The primary risk of aggregating healthcare data and it being stored in a nationally accessible database is one of security. If the nation’s entire healthcare records were accessible by those with the correct credentials, clearly an exceptionally high level of security would have to be put in place to ensure that the data could not be hacked or otherwise breached. We would also have to carefully consider who owns the healthcare data, and whether any healthcare providers are likely to be able to fund the massive investment required to digitise their data in such a way that it can be accessed universally.
How might you convince a traditional family doctor to subscribe to a new EMR system that will aggregate his data with that of many others?
I would look to convince the family doctor to subscribe to an EHR system - the name given to an EMR that focuses on the patient’s total health, allowing for the sharing of information with other healthcare providers. This would allow the doctor to readily gain access to lab results or specialist results, updated in real-time on the health record. The system would also likely be far better than their current EMR or paper records in terms of efficiency and ease of use, as the UI and UX of these systems improves over time. I would also stress the importance of switching to modern systems for their colleagues and the healthcare service as a whole - these new systems will allow clinicians and others to work together more easily than ever before, vastly improving patient outcomes and the working lives of all healthcare professionals.
In the future, how would you describe data aggregation and AI to a patient curious about what happens to their healthcare data?
I would explain that an AI system is one that is able to carry out a task that previously would have required human intelligence to perform. They draw on machine learning - the idea that a computer program can read and assess data, then adapt to new data without additional input from a human being. I would explain that through aggregating data, we can give the computer program enough information that over time it can learn more about the health of patients across the country, and in turn begin to make conclusions about the characteristics or treatments of certain diseases that would have been previously very difficult to make. I would explain that in order for the machine to work best, it must have access to as much information as possible, and that it will have no way of knowing who the patient is - i.e. that their data would be anonymised and used for the benefit of all.
What do you think the impact of AI will be on healthcare?
I believe that AI will have a huge impact on healthcare as we move through the 20th century. It will allow for the development of studies on a scale that has never before been seen, and with an efficiency that has never been possible before. AI will become more precise as it interacts with more training data, meaning that the scale of insight achieved will grow as more data is accessible. It will allow for more precise treatments, and more accurate diagnoses than ever before. It will also change the experience of doctors, allowing them to spend more time in direct patient contact and less time with electronic systems, reducing the incidence of burnout. It may also allow the better management of hospitals and staff, further reducing workload and improving patient outcomes.