Big Data: Rocking and Rolling Through Your Hospital

hospital-2Joe is a guest blogger for Solstice from MiGym, Sosltice's sister company focused on mobile health and wellness.

As an exercise, I'd like you to imagine yourself in Chicago in a familiar time of early August. Lollapalooza is starting near the end of the week, and temperatures are sweltering around the city. Luckily, (the fictional) Grant Park Hospital is prepared. Their hospital management system takes weather patterns, holidays and special events into effect when forecasting demand. As a result, hospital administrators have better information than ever to plan staffing and bed allocation. With a flexible floor plan system in place, the hospital is able to temporarily allocate more beds to its trauma center.

As expected, the steady stream of patients becomes a large influx. Because Grant Park Hospital is prepared, it can handle the load. The mobile phones that the EMT's carry listen to the surrounding conversation; the technology listens for certain terms, classifies the patient condition and scores the patient's mortality risk. Smart maps use current and historical transit patterns, patient's mortality risk, and hospital trauma level to most efficiently route patients to care.

At the Grant Park waiting room, extra staff have been allocated to handle the overflow. The extra staff don't need a desk; they check in patients using tablets. The tablets begin to learn about a higher volume of heat exhaustion and substance abuse patients, and automatically suggest those options to the staff. Using micro-presence technology, the nurse can see at any instant, from their tablet, which beds are empty and book the patient there.

At Grant Park Hospital, attendees and residents are allocated a list of patients to see, automatically prioritized using patient information like symptoms, vitals and history. As attendees visit each patient, the micro-presence system automatically brings up the patient record on the doctor's tablet. Based on the assumed affliction, the tablet can bring up a checklist of actions, frequently prescribed medicine, and possible drug interactions.

As the patient is being checked out, the hospital information system can score the patient's risk of readmittance and flag high-risk patients for extra scrutiny. If necessary, outpatient subjects may be equipped with a mobile vital monitor (like a mobile blood pressure cuff or ECG) that can immediately flag an on-call nurse or EMT from the patient's home if a dangerous reading occurs.

This vision of a healthcare future is closer to the next Lollapalooza than you think.  The anecdote above highlights the improvements that Big Data is making (and will continue to make!) in several areas:

  • Bed and staff allocation - Many hospitals are using bed and unit allocations from years ago. New analytics software can optimize allocations on a regular basis. In the future, these systems will take current allocation, seasonality and weather conditions into effect to better forecast patient demand on a day-by-day basis. Smaller, more mobile equipment and flexible floor plans can give hospital administration the ability to handle changing patient needs on a daily basis. Staff scheduling software can tweak projected resources based on expected demand. Hospitals can lower costs by not paying for unneeded resources; on the flip side, hospitals can improve patient care by better projecting staff demand early. Almost all hospitals call in nurses, or send them home, based on current workload. Better projections could eliminate a lot of this uncertainty.
  • Voice recognition -  Doctors spend about 40% of their time on a computer. Voice transcription could reduce or eliminate a large amount of the time that doctors and nurses spend entering notes. Most hospitals still use transcription services to handle doctor notes. However, now up to 43% of hospitals are rolling out some form of software voice transcription. In the future, these systems can be smarter. Like Apple's Siri, these systems will listen to the doctor's ongoing conversation and translate this to book tests and order prescription medication.
  • Smart suggestions - The other large data entry task for doctors are entering orders. Here, smart systems can take into account patient symptoms and diagnosis to suggest relevant tests and treatment. These systems can integrate with the patient medical record to present drug interactions immediately to the doctor.
  • Risk scoring -  Smart systems can begin to assess patient mortality risk to ensure that the most critical patients are being treated first. In addition, risk scoring algorithms can assess readmittance rates, which now directly affect Medicare reimbursement.
  • Outpatient monitoring - Outpatient system monitors can ensure that patients follow their treatment plans. Hospital monitoring systems could remotely track and report on blood pressure, glucose levels, and ECG readings. Smart prediction systems could scan these readings and report back to the hospital or EMT if the patient requires medical attention, or does not seem to be following their doctor's orders.

Improvements in mobile technology, smart sensors and learning algorithms can pave the way to a more efficient hospital with lower mortality rates. The best part of about this is that these systems can continue to improve. In the past, hospitals optimized operations by depending on clinical research, studies, medical literature, and in-house operational analysis. Modern learning algorithms, though, operate in real-time. With these new algorithms, hospitals can innovate and optimize at a much faster pace than was previously possible.

As these algorithms continue to operate in real-time, on massive real data sets, they will continue to get better. Like Apple's Siri, the voice recognition will continue to improve and become more accurate. Like IBM's Jeopardy champion system Watson, the information systems will get better at understanding and classifying patient disposition. Like Google Now, doctors' devices will get better at showing them the information they need, when they need it. And like quantitative trading systems used all over Wall Street, hospital monitoring systems will learn to automatically react to important vital signs.