From our buying habits to our banking, big data is transforming our approach to modern life. But the potential for revolutionizing healthcare is perhaps one of big data’s most tantalizing applications, and for obvious reasons—elixir of life, anyone?
Okay, so when we think about big data in healthcare, our minds may leap to a sci-fi future where all ailments are curable with nothing more than a quick injection in sickbay. But even in the real world, you’ll be surprised by the number of applications data analytics has in a modern healthcare system. From the run-of-the-mill to the seemingly magical, data analytics utilizes everything: from electronic medical records to genomic data, to improving healthcare for everyone. Perhaps that sci-fi future isn’t so far-fetched after all.
With this in mind, let’s explore nine fascinating use cases of data analytics in healthcare. We’ll look at:
- Disease diagnosis
- Disease prevention
- Electronic health records
- Improving clinical processes
- Informing treatment research
- Reducing hospital admissions
- Identifying appointment ‘no-shows’
- Minimizing health insurance risk
- Wrap-up and further reading
Ready to dive in? Then let’s go.
1. Disease diagnosis
Do you know what algorithms are great for? Pattern recognition. Classification. Prediction, too. Oh, and let’s not forget sifting huge amounts of data, and fast. These qualities have led to some pioneering machine learning algorithms that are revolutionizing pathology (the study of disease).
Pathologists are the first step in the diagnostic process. When doctors send off a tissue sample or blood test, it’s pathologists who analyze them to diagnose disease. This involves carefully examining each sample, which is no mean feat when they have hundreds to get through every day. Take into account human error, and even the best-qualified pathologist will sometimes make a mistake.
But thanks to data analytics, that’s changing. Certain companies—like PathAI and research groups like Google Brain (the eponymous company’s deep learning division)—are producing deep learning algorithms that can identify diseases from cancers to blindness in diabetics. By analyzing and classifying images, these algorithms can vastly reduce a pathologist’s workload. And they can improve accuracy by an astonishing degree. In 2019, for example, one tool correctly identified early-stage lung cancer in 94% of slides, outperforming six expert doctors. While doctors still get the final say, this highlights how humans and algorithms can work together in seamless symbiosis. More of this, please!
2. Disease prevention
Diagnosing existing diseases is one thing. What about predicting diseases before they even occur? Humans began to unlock this ability with the unprecedented sequencing of the first human genome back in 2003. But with every human genome containing roughly 100GB of data and companies like 23andMe now sequencing thousands every year, how can we possibly make sense of all these data? The answer—you guessed it—is with data analytics.
With so much genomic data now available to us, medical experts are developing sophisticated machine learning algorithms that can filter and interpret these genes to better understand how genetic diseases are encoded in our DNA. Comparing these data against anonymized medical records, scientists can now predict many heritable diseases ranging from hypothyroidism to gallstones, various cancers, and heart disease. While these insights don’t mean somebody will come down with one of these diseases, they can allow medical professionals to predict the possibility, offering patients early interventions and tailored treatments based on their unique DNA. Pretty cool, huh?
3. Electronic health records
As the world digitizes, so too does our medical data. Electronic Health Records (EHRs) contain a patient’s medical history, from allergies and diagnoses, to treatments and prescriptions. Okay, so we’ll admit, it’s still early days for EHRs. Not all countries have adopted them yet. But some, like the US, have poured billions of dollars into them, and others, such as Germany, are catching on fast.
The potential of EHRs is huge. With patient medical data stored in a single unified system, data analysts would gain unprecedented insights into human health. Predictive analytics could help us better grasp how risk factors like smoking, drinking, and obesity impact various diseases. EHRs would also allow doctors to tailor clinical treatments and search for hard-to-find information using natural language processing. They could also create data visualization dashboards that hone in both on emerging health issues, either for individual patients or for larger populations.
In short, EHRs have huge potential to transform medical care, tailoring it to a degree that’s currently unimaginable. Although we’re not there quite yet, there’s a lot of work going on in this area, so keep your eyes open!
While EHRs have the potential to transform healthcare, they also raise some (not unfounded) security concerns. Cyberattacks in the healthcare sector, which are increasing, have highlighted some major issues. For example, in 2017, the UK’s National Health Service (NHS) was brought to a standstill by a North Korean ransomware attack. Naturally, this wasn’t great news for the NHS or their patients. But it was made all the worse by the revelation that the attack was only possible because the NHS was still running Windows XP. Whoops!
This highlights just how important it is to stay ahead of the curve when keeping our data safe. As more of our data is digitized, the more opportunities there are for hackers to get their hands on it. This might seem less important when it comes to things like social media, but the privacy of health data is something altogether more sensitive. Luckily, data analysts are now being recruited in huge numbers to boost healthcare cybersecurity. Data analytics can help improve attack detection, manage potential risks and automate monitoring. This means we can predict attacks before they occur, alerting the system administrator to suspicious behavior. Looking for a job? Look no further!
5. Improving clinical processes
Hospitals are extraordinary places. They welcome most of us into the world. Many of us say our goodbyes in them, too. And between these mortal bookends, hospitals support us with everything from chronic disease to emergency care. Beneath all this, though, hospitals are in many ways no different from any other organization doing business. To run smoothly and to ensure quality care for their patients, they need effective systems in place. From seamless supply chain management to happy personnel, many factors contribute to how a hospital runs.
Enter healthcare business analysts. The role of business analytics in healthcare is a growing one. It can be useful for managing a hospital’s day-to-day operations (from patient waiting times to determining when machinery needs maintenance) and for higher-level strategy, too (such as improving the long-term capabilities of a particular department). Through accessing data on everything from clinical services to patient outcomes, healthcare analysts can identify previously undetected patterns in staff performance, patient care, and the overall way a hospital runs. This helps them find data-driven solutions that make things run even better. Granted, it’s not as sexy as ‘curing cancer’, but it’s still important!
6. Informing treatment research
Data analytics is big business in the pharmaceutical industry, especially for research and development (R&D). When successful, creating new drugs and treatments can be hugely profitable—but it requires a lot of time and financial investment, and it’s very high risk. One study found that only 9.6% of drugs entering phase I clinical trials will reach the market. Those aren’t great odds!
To reduce this risk, pharmaceutical analysts now use algorithms to search vast amounts of past research data. Obtaining insights from these data can inform their R&D effort, helping determine whether or not to proceed with new drug development.
It doesn’t stop there, though. Even once a pharmaceutical company has green-lit the development of a new treatment, they can use data analytics to speed up the process of clinical trials. For instance, by analyzing each participant’s past medical data, demographics, and the outcomes of past trials, they can highlight potentially costly flaws or areas for improvement. On top of this is the emergence of wearable devices that, in some cases, allow clinical trials to take place remotely. All these factors can lead to faster and more effective trials, further reducing the cost of bringing a new drug to market.
7. Reducing hospital readmissions
The unexpected readmission of discharged patients is one of the most costly aspects of running a healthcare system. It’s a common problem. In the U.S. alone, hospital readmissions cost $26 billion every year—but this is one problem the healthcare industry is already trying to solve.
One emerging tactic is the development of patient risk scores. Using demographic and diagnostic data, these assign each patient a number. This number tells the healthcare system how much a patient is likely to cost compared to the average. While transforming individuals into statistics might sound mildly dystopian, it’s important to note that it doesn’t affect a patient’s treatment. Or rather, it does, but it helps improve their care—using patient risk scores, medics can develop and offer tailored screenings and interventions, minimizing the risk (and cost) of later readmission. In turn, these costs go back into the system, supporting those who need it most.
Another tactic is the use of real-time push notifications. These remind patients to take their medication or to keep on track with a healthy living plan. While both these solutions are still evolving, they’re already proving potent tactics. Expect to see more of this in the future!
8. Reducing appointment ‘no-shows’
In the U.S., missed medical appointments cost an estimated $150 billion per year (and you thought hospital readmissions were pricy!). Reducing missed appointments, then, is a significant way for hospitals to reduce costs. Fortunately, there are already researchers looking into ways of doing this. And do you know what tool they’re using? Yep, you guessed it: data analytics to the rescue once again.
Using predictive modeling, a study in 2018 analyzed data from over two million outpatient appointments across 14 different sites and 55 clinics. By applying logistic regression models, the researchers were able to identify 4,819 patient no-shows. Interestingly, they also found that clinic-specific data led to more accurate predictions, even though this meant far less training data for the algorithm. This highlights the notion of ‘garbage in, garbage out’. Namely, the quality of data is far more critical than the quantity. While this was a preliminary study and didn’t focus on cost savings, insights like these would undoubtedly allow hospitals and clinics to operate more efficiently and make better use of staff time.
9. Minimizing health insurance risk
In one way or another, all the use cases in this post rely heavily on predictive analytics—and there are few areas more reliant on this than the health insurance industry. Attempting to forecast the future is nothing new for insurance companies: actuaries have long been gauging applicant risk, considering factors like their profession, location, age, sex, and past insurance claims. Even where they shop.
However, with the arrival of big data and the surge of eager new analysts, health insurers can now price their health plans with much higher degrees of accuracy. Rather than using data to blanket refuse to insure people (as was once common) they now increasingly use tailored insurance models. In short, they can use unprecedented data insights to determine which health plan is best suited to an applicant’s lifestyle and risks. This means accepting more applicants, which is good news for all parties involved. Everyone wins!
10. Wrap up and further reading
Data analytics is a powerful tool, a fact that is especially evident when we look at the healthcare industry. In this post, we’ve explored some of the emerging ways in which data analytics underpins our approach to health. From improving clinical care to streamlining operations, revolutionizing disease diagnosis and pharmaceutical research, there’s something in the healthcare industry for everyone.
Fascinated by the potential of deep learning and data analytics in healthcare? Perhaps you’re a scientist hoping to create new drugs—or maybe you’re obsessed with improving processes? Wherever your skills and interests lie, learn more about data analytics with this free, 5-day data analytics short course. You can also read the following introductory topics to learn more: