Do you have the right tools to get the best out of your data science team?
You have plenty of data, a solid IT infrastructure, and a team of data scientists on board. Still struggling to streamline your data processes? You are not alone. Read on to find out how EdgeLeap helped the data team of UMCU Psychiatry department to set up a sustainable and collaborative data science environment and bring data-driven insights closer to the bedside.
Linking of different data and information helps to unravel complex interactions and obtain valuable insights to support decision making. Floor Scheepers, professor Innovation in Mental Health Care, University Medical Centre Utrecht (UMCU) recognizes the value of leveraging data science technologies to decipher complex interactions between the environment, genetic vulnerability and physiological functionality to improve psychiatry diagnoses and optimize patient care.
“By linking internal patient and organisation data we can discover patterns in how different factors affect patient behaviour and wellbeing” says prof. Scheepers. “This helps us to predict and anticipate to events at our department, such as knowing which patients are more likely to be aggressive under specific circumstances or who will have an unusual reaction to medication. If we can timely prevent incidents by simple intervention – such as taking a patient at risk for a walk – this has a huge benefit both for the patient and the caretaker.”
To realize the vision of improving patient care by leveraging advances in data science, prof. Scheepers has brought together a big data team to support needs of the Psychiatry department. The team’s task was to analyse collected hospital data to answer questions of caretakers and help them obtain insights they were looking for. With the expertise of individual team members, availability of the UMCU RDP (Research Data Platform) infrastructure, and access to enterprise statistical software, the team had all ingredients to succeed. Still, the data process did not run smoothly. Routine tasks such as data transformation, files upload / download, and timely updates were performed ad-hoc and manually, which was both time consuming and error-prone. Moreover, scientist’s favourite tooling was impossible to use in the workflows that ran within the hospital infrastructure. Data scientists had preference for different programming languages, typically incompatible with the standard enterprise software, hampering collaboration and reuse of code. The data process was in need of a reform.
‘This platform forms a foundation for a sustainable data process.’
This is when EdgeLeap stepped in. “Our approach was to first carefully listen to needs of all stakeholders: data scientists, clinicians, nurses, and UMCU IT professionals.” says Marijana Radonjic, co-founder and CEO of EdgeLeap. “By matching technology possibilities with the user needs, while complying to the organizational constraints, we could propose a technology solution to optimally support the Psychiatry big data team within the UMC environment.”
EdgeLeap implemented a data processing platform enabling the team to organise their current way of working in an efficient, sustainable and collaborative environment. Based on existing modules for code sharing, organization, and pipeline automation, EdgeLeap has implemented a platform to help the team to share, reuse, document, and automate their data processes. Radonjic explains: “The platform allows the Psychiatry big data team to build upon the organization-wide infrastructure and build sustainable data processing pipelines, while gaining the freedom to combine and experiment with different datasets, tools, and algorithms. This approach bridges the highly structured, but rigid hospital-wide data processes with the day-to-day needs of a specific team focusing on a single line of care.”
Overview of the data processing platform. The platform enables automatic scheduling, sharing and documenting of the data pipelines. This automatizes data process and safeguards reproducibility and transparency of the code, while providing data scientists with flexibility in their way of working.
To bring the data-driven insights closer to the bedside, a series of automatised infographics reports have been generated to provide caretakers with continuous insight in the current situation at their department. Such explicit display of what is all possible with data aims to inform and engage caretakers. “We hope to stimulate our end users to ask even more questions and collect more (accurate) information – which in turn reinforces the entire data process and sustainably captures the value of data analytics”, adds prof. Scheepers.
Example of the automatized infographics report displaying status of a care division at UMCU Psychiatry department. The reports are periodically scheduled and generated automatically based on existing data in the hospital database.
‘We hope to stimulate caretakers to ask even more questions.’
“This platform forms a foundation for a sustainable data process at the Psychiatry big data team. The flexible, modular design allows the platform to grow and change along with requirements of future team members and new data resources.” stresses Radonjic. “The transparent coding process fosters cross-learning and building upon the legacy of others. This minimizes risk of losing knowledge, which is becoming critical in today’s rapidly changing working environments.” The implemented solution also boosts the data science culture within the team, creating a more attractive environment for modern data scientists. She adds: “Repetitive, time consuming tasks are now largely automatized, freeing time to tackle exciting data science problems. The team members now have freedom to use tools and programming languages of their choice in an interoperable and collaborative way. This is a prerequisite for getting the most of their time and talent.”
Prof. Scheepers concludes: “Working with EdgeLeap was really helping and created a positive vibe in our organization. They turned out to be a good listener to our problems, very open minded and creative, and brought solutions that were realistic and in line with our vision.”
Interested in hearing more about this topic? EdgeLeap and UMCU are organizing the “Meaningful data & smart fashion” session at the DRIVE festival (25-26 October 2017, Eindhoven, The Netherlands ) to steer a discussion on how to leverage innovations in data science to make faster and better informed decisions in healthcare. More information: