Jenny Zou
Data Scientist
IHiS
Joined in 2017

Using Data for Good

In the thick of the COVID-19 pandemic, Jenny Zou and her Data Analytics and Artificial Intelligence (DnA) team at IHiS worked with Changi General Hospital (CGH) to build an AI predictive model.

Known as the Community Acquired Pneumonia and COVID-19 AI Predictive Engine (CAPE), it can help doctors quickly detect whether patients have mild or severe pneumonia.

“As pneumonia is one of the key symptoms of deterioration among COVID-19 patients, the model enables timely triaging and treatment which in turn mitigates their likelihood of being admitted into the intensive care unit,” explains the 32-year-old data scientist.

The Emergency Department and ward doctors at CGH can be given early warning should there be deterioration in patients’ conditions and prescribe appropriate measures to improve medical outcomes.

Most recently, Jenny and her team enhanced CAPE by developing an “image explainability” algorithm named Ensemble XAI to help clinicians make more accurate X-ray image interpretations and diagnosis. The Ensemble XAI benchmarked well against industry standards and has gained the endorsement of a panel of expert radiologists.

The work done here is one of Jenny’s proudest moments at IHiS to date, particularly for someone who graduated with a Mathematics degree and didn’t expect to embark on a HealthTech career.

To cap it off, she collaborated with her colleagues to write two papers about CAPE, which were published in the Association for Computing Machinery (ACM) Special Interest Group on Knowledge Discovery and Data Mining (SIGKDD) - a leading publication on data mining and analysis as well as leading medical research journal BMJ.

Making an Impact

Prior to CAPE, Jenny had worked on a project with the National Dental Centre Singapore (NDCS) and KK Women’s and Children’s Hospital (KKH).

Patients who have registered for medical appointments yet not show up is a longstanding issue that clinics have had to face.

“As a result, frequent patient no-shows put significant strain on clinical operations and is an inefficient use of limited hospital manpower and resources,” explains Jenny. “With growing demand at outpatient clinics, it was increasingly necessary for healthcare institutions to optimise staff allocation and clinic workloads as well as to improve patient experience.”

With this challenge in mind, Jenny and her team got to work.

At NDCS and KKH, they deployed an AI-powered No-Show Predictive Model that identified patients who were likely no-shows based on their risk scores. This was to better optimise the centre’s resources, and reallocate appointments to other patients who needed to see the doctor earlier.

About one year’s worth of data, consisting about 3 million records, was used to develop the prediction model. Another year’s data was used to validate the model, which had a prediction accuracy of 77 per cent.