“Predictive algorithms have potential to drastically improve hospital efficiency”

Mentioned Professor Svetha Venkatesh, Director, Centre for Pattern Recognition and Data Analytics (PRaDA), Deakin University, Australia who shared her thoughts on the importance of data analytics for healthcare management in India

Svetha Venkatesh is the Alfred Deakin Professor who along with her team has tackled a wide range of problems of societal significance, including the critical areas of autism, security and aged care. The outcomes have impacted the community and evolved into publications, patents, tools and spin-off companies. This includes 529 publications and three award winning start-up companies: Virtual Observer, iCetana, and TOBY Playpad.  

The Centre for Pattern Recognition and Data Analytics (PRaDA) of Deakin University, Australia has been using data insights to address real-world problems through efficient usage of Electronic Medical Records (EMR), which would transform the healthcare system. In a freewheeling conversation with the BioVoice News, Professor Svetha Venkatesh, Director, PRaDA shared her insights on the potential of EMR systems in tackling various health related issues besides the details of her partnership with Max Healthcare. Read the excerpts:  

BV_icon-150x150How old is the Electronic Medical Records (EMR) system and how do we maintain the standards? 

Electronic Medical Records are not new but fairly recent. These are the wealth of information and essentially the journey of people through hospitals. The patients come to the hospitals, and then what happens to them there, their experiences and what kind of treatment they undergo. It is basically the journey of patients i.e. arrival, diagnosis and treatment which gets recorded electronically.

There is a world agency that regulates the standard of these medical records and currently is called International Classification of Disease-10, which is the tenth revised version. Therefore, the EHR software have to ensure compliance to ICG-10.

BV_icon-150x150How do predictive algorithms work and how can they improve the quality and efficiency in clinical procedures?

Predictive algorithms have huge potential to drastically improve hospital efficiency. The interesting part is that medical data is not recorded in the text but a code which adds huge value for the machines. The codified data that is ready with information is then put to use by the machines for predictive analysis.

The predicative potential of EMR is therefore huge for clinical procedures. You are going to have the patient’s journey across years and they can start asking questions as well as find the answers. For these questions, one earlier had to do experiments and clinical trials to understand what is happening in the population of your interest.

Recently Prof Venkatesh has been awarded an ARC Laureate Fellowship by the Australian Research Council (ARC). The Fellowship will provide over $3 M in funding to support a project that aims to determine how pattern recognition can be harnessed to accelerate and expand the capability of experimental optimisation that underpins scientific innovation.

BV_icon-150x150How can such tools help in assessing the effectiveness of a country’s health system?

Since the predictive capacity is huge, the large scale utilization can give good results. These predictive algorithms work on the basis of a pattern of data. If we have a certain population under our radar, we have to try and find the maximum data. The maximum combinations of that data help in the predictability to get better. The more data you have, more effective the predictability model.

What can be it used like is, for example if a patient has to be discharged, one can predict whether he will come back or not within 30 days of attending the hospital. The prediction of adverse events or many types of hospital events can help in improving hospitals’ efficiency. This is just an example.

BV_icon-150x150Is the data refined? Do hospitals really leverage the raw data enough to get generate insights. What should be done?

Currently the data is just wild and there is no categorization. The hospitals don’t pool data. The moment they pool in the data, it will become more precise and allow the algorithms to do a better job.

Depending on the patient profile, sometimes the patients come and leave. The data utilization could help the hospitals to take decisions regarding the patient health and treatment. The predictive analytic models are very good with sometimes 60-70 percent accuracy and even 80 to 90 percent. More data you have, more accurate the predictions you make.

BV_icon-150x150What is the deeper aim behind partnership on healthcare data mining between Deakin University and Max Healthcare?

Deakin University, Centre for Pattern Recognition and Data Analytics in Australia signed an agreement with Max Healthcare to develop a project that will focus on data analytics for healthcare management in India in 2015. Max Healthcare runs 10 hospital branches in India, of which eight are in the National Capital Region (NCR). This is the first of its kind venture, where both partners have put big data to work across a large array of medical records including admissions, diagnosis and computerised registries with the aim of identifying critical safety issues and assisting clinical efficiencies.

This project addresses this pressing need, leveraging state-of-art and verified techniques in data analytics to inform clinical decisions. The outcomes of this project are critically important, from economic, patient safety and systems perspectives.

BV_icon-150x150Please share the details of the project? How far has the project progressed, so far? What are the specific outcomes?

We are replicating the risk analysis model we had tried earlier in Australia. There the question was whether the patient is going to return to hospital in 30 day and based on risk analysis if it was yes, then there was no reason to discharge. The risk determines if the person requires longer medical care so that he or she doesn’t need to come back. We identified the risk problems and did the predictions.

Here in India, the immediate project focus is on heart disease, specifically on patients with symptoms of Acute Myocardial Infarction (AMI). The primary objective of the project is to search through the existing data sets for hidden patterns of both the predictable and preventable events in managing the healthcare of individuals. The focus of the current work is to identify patients at risk of frequent unplanned re-admissions.

Frequent unplanned re-admissions diminish the patient care and institutional performance and adds to the cost of managing AMI patients. Recent observations indicate that out of five admitted AMI patients nearly one patient is readmitted within 0-30 days after discharge. To reduce unplanned readmission, it is mandatory to recognize the high-risk patient during initial admission and then identify the risk factors and administer medication accordingly.

BV_icon-150x150Has the data been really used to its potential? What about the data security and privacy? Are there any laws?

Currently the world is experiencing a data deluge. We are living in the world of big data where the traditional methods have failed to manage this scale and complexity. At the centre of this data-wealth, is the health sector with data on admissions, diagnosis and outcomes, spanning a bewildering and disconnected web of images, computerized records and registries. There are no systems to manage this big data. Despite the wealth of health solutions trapped in the ore, this critical resource is not mined. The result is write only data, mostly unused, with untapped potential to identify critical safety issues, as well as service and clinical efficiencies.

The security of data is indeed an issue. While the general data is always at the risk, the mined data has to be protected more because of vulnerability. There are no defined laws as the area is still developing. Institutional bioethics committees generally monitor the issues on case to case basis at various organizations where the work related to data or big data related to clinical procedures or health is happening.