Health Care

‘Data science helps to find a needle in a haystack’

Esther de Vries

Coordinator Data Science

A patient is an individual, not an average. However, numbers can make health care more personal, Esther de Vries argues. She is coordinator data science at the Jeroen Bosch Hospital in ‘s Hertogenbosch as well as pediatrician and endowed professor at Tilburg University. To optimally take advantage of the opportunities data science offers, the hospital has to adjust its organization and methods. Pipple is advising the hospital on this.

Data science to personalize healthcare

The Jeroen Bosch Hospital highly values continuous improvement of patient care, in which ‘personalized healthcare’ is an instrumental element. Personalized healthcare is care, which is not just based on the average of people with a similar syndrome, but specifically focuses on the needs of the individual patient. Esther: “To base predictions on an average is not wrong; it’s just not enough. Everyone is different.”

A lot of data required

“Data science helps to find a needle in a haystack more easily”, Esther says. “You take patients who are diagnosed with a specific illness or side effect of a medicine. In their electronic patient files, data concerning the origin can be extracted. You compare these data with the data from patients without the illness or side effect. In this way, you are able to more accurately predict illnesses or side effects, provided that the research population is sufficiently large.”

Electronic patient file is priceless

“Files are mainly used as a means to archive, while they can be a giant data source to improve healthcare.”

– Esther de Vries – coordinator data science at the Jeroen Bosch Hospital

Although nearly all data in healthcare are digital since the introduction of the electronic patient file, data science now reveals itself as a new discipline to hospitals. “Our professionals are educated in health. They are not particularly handy with computers. Due to that, files are mainly used as a means to archive, while they can be a giant data source to improve healthcare.”

Standardization is difficult

Unfortunately, the usefulness of the data is often limited, Esther notes. “Doctors and nurses still report in the same way as they did in the paper era. There is a lot of free text. As a consequence, for example, many different words are used for the same phenomenon.” Standardization of reports would be a huge improvement. “We have been working on that for several years, but it is still difficult. Staff lack time to change their way of reporting.” To utilize free text nonetheless, the Jeroen Bosch Hospital is also learning to exploit the possibilities of text mining.

Pipple handles it in a practical way

To develop data science within scientific research, the Jeroen Bosch Hospital has approached Pipple for aid in this. Esther: “We want to be able to gain knowledge from the data ourselves and to develop algorithms that predict the chance of the occurrence of rarities.” Pipple investigates what is required in the fields of staff, organization, hardware, software and logistics in order to attain these objectives. Their approach is practical. “Pipple starts with cases that we have selected together. During the process, they encounter the restrictions of the current situation. In this way, they collect all limitations and think of possible solutions.” Furthermore, Pipple explores the possibilities to present data analyses in a way that is understandable for doctors and patients. They incorporate their solutions into a pragmatic plan. This is very important to Esther, after all: “The hectic pace of a hospital is, and remains, a given fact.”