Delwyn Armstrong | Head of Analytics
I was lucky enough to attend the Health Information and Management Systems (HIMSS) conference in March held in Las Vegas – the largest health informatics conference in the world – 45,000 attendees!
My prime interest was in bringing home practical data analytics ideas and gauging the future direction for systems and analytics. While previous informatics conferences have focused on new applications, with analytics as a side-line (albeit a growing one), analytics was centre stage here.
The keynote speaker, Eric Schmidt, former CEO of Google, offered the formula for driving a step-change in improvement of care: DATA + CLOUD + NEURAL NETWORKS + REINFORCEMENT LEARNING.
An explosion of data coupled with machine learning (neural networks) is creating a revolution in IT now – however, he says, health is lagging other industries. He also stressed that cloud platforms are safer and easier to use than on premise platforms, and urged public cloud adoption.
Data analytics, especially predictive analytics, is gaining momentum as US healthcare moves into a post-EHR (electronic health record) age. That is, acquisition of an EHR is so common now that the conversation has moved on to leveraging the value of the data collected. I was pleased to see not only development of predictive algorithms but an emphasis on how to successfully implement them in real clinical workflows. In fact, it was stated that the most significant driver for improvement in care outcomes was access to real time data for care-related decisions.
One of my favourite presentations was ‘Stacking Predictive Models to Reduce Readmissions’ by UnityPoint Health, Iowa. They not only changed my understanding of readmission patterns (for example, some patients are more likely to readmit in the first few days, while others are more likely to deteriorate over 3-4 weeks then readmit), but also explained their multi-faceted approach – providing real-time information to their inpatient and chronic disease navigators, and embedding readmission risk assessment in daily huddles.
HIMSS showed predictive analytics used to reduce length of stay, detect paediatric sepsis, improve hospital capacity management, develop a congestive heart failure checklist and reduce infection rates.
The new generation data visualisation tools appear to be routinely used in American healthcare, with Qlik and Tableau being the most popular, so we are not lagging behind there! (We’ve recently implemented Qlik Sense at Waitemata, you see).
I was half looking for some fear and loathing in Las Vegas; I left with a fear of under-utilising our growing data goldmines and a loathing of the smoking indoors acceptability, but energised by the opportunities to contribute to better healthcare through data.