COVID-19 has caused rapid evolution in the field of healthcare technology and these advancements have the potential to shift how care happens quite radically in the years to come. From new vaccine techniques to robotic surgeries, the pandemic has driven the development of many new technologies.
One arena that has received significant attention since the start of the pandemic is predictive analytics. These tools have helped relieve some of the uncertainty with the pandemic, such as a mortality risk calculator from the Johns Hopkins Bloomberg School of Public Health. This tool has helped inform treatment decisions and even guided vaccination rollouts.
Predictive analysis has many applications in the field of healthcare and will likely be applied in several ways even long after the pandemic has subsided. Some of the applications of predictive analysis that have already been explored include:
1. Staying ahead of equipment maintenance.
Many types of medical equipment degrade over time and with regular use. When these machines experience errors, there can be significant delays in diagnosis and treatment. Predictive analytics can be used to identify what proper maintenance intervals can be implemented to avoid unanticipated downtime.
If hospitals and clinics know when a part is due to be replaced, they can plan maintenance during normal downtime or figure out another solution to minimize unscheduled disruptions to workflow. This technique has already been employed in MRI scanners. These scanners can relay technical data to a monitoring system that analyzes them and gives warnings of impending technical issues. In the future, many different pieces of medical equipment will likely be connected to the network and relaying this sort of data to uncover maintenance needs in real time.
2. Providing personalized care for palliative patients.
Penn Medicine has developed a system called Palliative Connect that collects information about patients from the health record association with the hospital system. Then, the program generates a prognosis score based on dozens of factors to figure out likelihood of survival for the next six months.
So far, this system has helped the palliative care team recognize patterns associated with poor prognoses and be more proactive in its approach. In its pilot run, the program identified 85 patients for consultation with the palliative team.
During the same period, only 22 would have been identified for consultation without the analytics. This sort of algorithm can help keep care personalized for patients most at risk and ensure they get connected to the ideal type of treatments.
3. Reading imaging studies quickly and accurately.
Researchers at Stanford University have been using predictive analytics to screen chest x-rays in a matter of seconds. This sort of technology can help in the diagnoses of patients in the urgent care or emergency settings when time is of the essence.
While the algorithm has not yet been introduced into the clinical setting, it could help triage results while waiting for formal review from a radiologist. When an abnormality is detected by the predictive analytics, the result can get flagged for immediate attention from radiologists instead of giving preference to presumably normal studies.
Predictive analytics could be applied to a wide range of other medical imaging modalities. Already, experts are figuring out how to use this technology to improve cancer care, such as in identifying lymph node metastases more quickly. While these results do not replace physician readings, they can augment their value.
4. Identifying patients who are deteriorating.
For patients with COVID-19, predictive analytics helped estimate morbidity and mortality for critical patients. Similar algorithms can be employed in the ICU setting and general wards to identify patients who are deteriorating earlier than would be evident clinically.
These predictive algorithms use many different information points, such as vital signs, to find patterns indicative of the early signs of deterioration. With this information, providers can intervene before the situation gets worse and avert potentially disastrous outcomes.
Also, the technology has been used to calculate risk of death or re-admission within a 48-hour period, which can help inform decisions about whether or not it is time to transfer an ICU patient to a general medicine floor or even discharge that individual from the hospital. These technologies have largely been developed for use in a virtual ICU setting to allow providers to monitor remotely, but will have many applications even when ICUs are less crowded.
5. Keeping patients out of the hospital.
The value of predictive analytics extends beyond the clinic and hospital settings. When patients get discharged from the hospital, they generally do not have much long-term health monitoring. This leaves them at risk for re-admissions that could have been avoided.
Systems have been developed that take data from electronic health records, fall detection pendants, and other medical alert services to identify people who are at risk of injury or illness. With this information, providers can reach out before a problem occurs and be proactive with preventative measures.
This sort of technology could significantly reduce the costs associated with acute care and rehabilitation. Moving forward, this technology could become more sophisticated and be tailored to each person’s particular needs. Physicians could be able to set parameters for check-ins based on patient history and prior admissions to the hospital.