December 10, 2019
Table of contents
With the increasing complexity and variety of data generated within the healthcare field, it is important to employ tools that can help turn data into actionable insights. Business intelligence (BI) is exactly this type of tool.
Turning to business intelligence consulting, you will be able to unlock the full potential of data accumulated within and beyond your organization. BI based on big data contributes to patient treatment by facilitating triage, predicting the risks patients are likely to face, avoiding readmissions, and improving patient satisfaction with care delivery.
BI is particularly relevant now with the centralization of electronic health records (EHR), the popularity of social media as an unlikely source of healthcare data, and the relative affordability of the technology required for BI implementation.
Healthcare BI aggregates and analyzes different types of healthcare data, including such traditional records as medical history and financial data, along with unstructured data gathered from around the web and the internet of medical things. This data is not limited to care providers’ settings but can come from outside their organizations, too.
When you look at the mortality indicators or the prevalence rates for heart disease and diabetes, those are related to so many circumstances that occur outside of the traditional healthcare system.
PhD, Director of the Centers for Medicare and Medicaid Services Office of Minority Health
BI helps to understand the state of your healthcare organization and make informed decisions regarding potential opportunities for growth and improved services. BI spans three categories:
Visualization tools are very useful as they help extract the necessary information from the heap of data when needed. For example, before accepting a patient for an appointment, the physician can quickly access the relevant data in an easy-to-view format.
Business intelligence is about performing different types of analytics on huge, often unstructured, datasets. It covers data mining, big data analytics, querying, modeling (including that based on machine learning), and reporting.
BI supports healthcare professionals with the following types of analytics:
Business intelligence relies heavily on data. The more data an organization can aggregate, the more accurate outcomes BI delivers. Therefore, the first step toward adopting BI is to aggregate as much relevant data as you can get, and adopt the technology for medical staff to capture the data as they encounter it.
If you have better technology at the point of care, you can get better data in a more useful form.
Head of Health and Private Sector, KPMG
BI can contribute to patient satisfaction in several ways. It helps predict a patient’s needs so that the healthcare facility is prepared to act accordingly. It also helps produce customized treatments and contributes to care delivery optimization.
BI can analyze available patient data and visualize it in a way that helps to understand the situation of interest. For example, the graph below sums up appointment-related information and patients’ satisfaction levels.
The University of Pittsburgh Medical Center adopted business intelligence to predict the impact of the flu season on patients. Their BI predicted which patients will be needing more attention, and who will require emergency care. Additionally, the Medical Center was able to forecast readmission and take preventive measures, thereby avoiding regulatory fines and improving their outcomes.
Every doctor can attest that there is no one-size-fits-all treatment, and most treatments are never the same for different patients. The same medicine might work well on some patients but do nothing for others. It would be impossible for a healthcare practitioner to sift through all the relevant patient data manually to come up with a personalized treatment.
BI can analyze huge data sets and extract hidden correlations automatically. This can help practitioners customize treatments based on every patient’s specific needs.
For example, predictive analytics can be used to predict fatality during surgery based on the patient’s current condition, medical history and treatment. If there is a high risk of fatality, an alternative treatment will be presented.
Patients’ satisfaction is not limited to the treatment they receive. It is rather a combination of different factors including the transparency of care, communication with the caregiver, as well as empathy and respect shown.
As caregivers have a great influence on patient treatment and satisfaction, BI can become the tool to evaluate caregivers individually and collectively. This can be done by analyzing not only surveys filled in by patients but also social media sentiment and other accessible data.
However, the goal here is not to stress caregivers and make them constantly monitor their own performance. This is more to give an indication of the level of care offered by a healthcare facility, and to help react immediately in case of inappropriately low levels of care.
One of the hardest processes in healthcare is triaging incoming patients and deciding whom to admit and whom it is safe to discharge.
Unnecessary hospitalization leads to the wasted time of medical emergency staff and hospital beds taken from those who might have needed hospitalization more. Additionally, it results in unnecessary expenses borne by patients, insurers, and hospitals. At the same time, if the staff fails to admit a patient in need of monitoring, this can result in complications and even death.
Business intelligence tools geared for predictive analytics, along with software solutions for healthcare business process management, have the ability to rank patients’ cases and thus assist in triage.
NorthShore University HealthSystem operates four hospitals in Illinois. Their emergency staff has always found it challenging to decide on the admission and monitoring time for patients with chest pain. Chest pain is a serious problem that might be an indicator of a heart attack. However, it can also be a symptom of a less serious condition such as heartburn. Doctors are very cautious when ruling out serious conditions and discharging patients with chest pain, and would rather keep them under monitoring for a longer time.
To minimize those long but often unnecessary stays, NorthShore opted for experimenting with predictive analytics. They designed a system that could assign a “heart ranking” to every incoming patient with chest pain based on the HEART score first trialed in the Netherlands in 2017. To develop a predictive model for their setting, NorthShore composed a cross-functional team including the hospital president, two emergency physicians, and nursing lead among others.
The algorithm at the core of the system assigns a “heart score” to every incoming patient. This ranking is based on a patient’s history, electrocardiogram, age, risk factors, and initial troponin. As a result of this trial, patient monitoring days dropped by 10%. At the same time, following the system’s prescriptions did not increase the number of readmissions.
Using BI-based predictive analytics, it is possible to forecast which patients are at a bigger risk of catching or developing a disease. This allows earlier intervention and prevention before a condition becomes too serious.
According to the Society of Actuaries report, nearly 75% of healthcare expenses are coming from only 17% of patients. Being able to identify patients at risk and take preventive measures is what can eventually reduce treatment costs.
When we think about the uses of data and analytics in the value-based environment, the theme is trying to swim upstream to get to prevention.
Chief Medical Officer, Atruis Health
Traditionally, the risk could be forecast using factors such as blood pressure, glucose level, age, and medical history. Now, thanks to the development of fitness wearables and other connected devices, it is possible to incorporate factors such as diet, health habits, and employment conditions, and to monitor patients in real time.
Virginia-based Carilion Clinic employs predictive analytics to spot patients prone to heart diseases. This tool was tested on 8,500 patients and had an impressive 85% success in predicting heart failure, which improved patient care and saved money. The tool analyzes traditional structured data in addition to unstructured data such as doctors’ notes.
In another example, the University of Pennsylvania developed a predictive analytics tool that uses EHR data to predict which patients are about to develop sepsis 12 hours in advance.
Unplanned hospital readmissions tend to be the costliest services in healthcare. Readmissions affect patient satisfaction and increase the risk of negative outcomes.
Many factors can contribute to readmission, and it is not always about incorrect treatment. Patients can fail to schedule a follow-up appointment, may not afford medication, or just fail to follow lifestyle recommendations.
BI tools can take all those factors into account when predicting risks and helping resolve them without the need for readmission. For example, UnityPoint Health in Iowa has been using predictive modeling in a three-year pilot project. As a result, their readmission rate dropped by 40%.
Patricia Newland, a family physician, experimented with a BI tool to prevent readmission of one of her patients. The tool predicted that the patient would experience a particular set of symptoms within the next 13-18 days. Patricia asked the patient to contact her immediately if the patient experienced those symptoms, while the physician left a window for an appointment in her schedule to deal with the situation immediately and prevent escalation.
The patient indeed experienced the symptoms within the predicted timeframe. She contacted Patricia and had an appointment on the same day, having her symptoms treated swiftly and avoiding readmission.
Healthcare facilities are being driven toward opening their data and making it available for research purposes. In this scenario, BI can be used to analyze large volumes of health data including medical conditions and demographics to generate cohort health patterns. This effort calls for healthcare automation via RPA or other technologies.
These patterns can help healthcare organizations better target their population health campaigns, such as those against obesity. BI also plays an important role in epidemiology, as it can predict outbreaks and alert healthcare facilities to them before they spiral out of control.
Another application of BI in cohort treatment is clustering cancer patients with similar characteristics. Analyzing their historical data helps predict how their conditions will be developing in the future.
BI opens valuable opportunities for healthcare organizations to improve care quality and increase patient satisfaction. However, BI benefits are not limited to interactions with patients but include hospital operations, supply chain management, and staff allocation, to mention a few areas.
While bringing many benefits, BI poses new risks that healthcare organizations need to be aware of:
Some leaders believe that their organizations do not have enough data to build up BI capabilities. Don’t let that stop you. Predictive models can be created with less-than-perfect data. With all the possibilities BI has to offer, it can bring you to a new level of care quality and performance.
Medical data analytics solutions by Itransition drive valuable insights from a pool of health data and help improve care and performance. Get a free quote now.
We reflect on the most up-to-date applications of big data in medicine across the care cycle and explore the most relevant know-hows of the technology.
Learn how the internet of medical things can increase the satisfaction of your patients, improve internal processes and staff allocation.
Read about Itransition’s 10+ years long cooperation with a US-based multinational to create their flagship pharmaceutical data analytics products.
Itransition helped the customer bring their solutions up to speed and further secure their leading position in the healthcare analytics app market.
Find out how machine learning personalizes patient care, alleviates pressure off doctors, and opens opportunities for innovative medical research.
EHR implementation is a complex process loaded with challenges. We present some tips to smoothen your EHR deployment. Have a look.