It was on 26 June 2000 that the entire human genome was sequenced. “We have caught the first glimpse of our own instruction book, previously known only to God,” said Dr Francis Collins, one of the scientists leading the Human Genome Project.

That was a revolutionary step for healthcare. Analysing genomic data can allow scientists to predict one’s likelihood of suffering from a disease – such as the risk of Down syndrome in pregnancy.

Cindy Maike, Vice President, Industry Solutions & Value Management of Cloudera, explores how healthcare providers can use AI to unlock new possibilities in healthcare.

AI for genomics

With genomic data, clinicians can study how genetic differences contribute to complex diseases like cancer and diabetes. This will enable doctors to tailor treatments to a patient’s genetic makeup, otherwise known as precision medicine.

Singapore’s healthtech agency IHiS is planning to expand its use of AI into genomics and medical imaging, its Director of Data Analytics & AI told GovInsider.

The use of AI in genomics will be helpful in Covid-19 treatment, Maike says. Studying a person’s genes can help healthcare providers understand how various virus strains will infect people in different ways and their potential long-term effects.

But the human genome is massive and makes up about 300GB to 1TB of data, according to intelligence company CB Insights. AI will be useful in analysing this data, and identifying patterns that clinicians won’t be able to spot.

The US-based Inova Translational Medicine Institute aims to deliver precision medicine, but previously faced challenges of integrating and analysing vast amounts of patient and genomic data.

It worked with Cloudera to integrate these massive datasets into a single data warehouse hosted on the cloud. This has helped clinicians access and analyse data to test new theories and uncover new patterns.

Tackling AI bias

AI holds immense potential for good, but healthcare providers must also address concerns of bias and privacy, Maike says.

AI makes decisions based on data – which can include social inequalities or existing biases. A 2019 study revealed that an algorithm used in US hospitals to allocate healthcare had been discriminating against Black patients, Nature reported.

To ensure AI produces reliable and ethical results, IHiS verifies results against real-world outcomes before a model is deployed. It also plans to conduct pilots and reviews to uncover any potential issues, said its Director of Data Analytics & AI.

Running multiple simulations can help to test the reliability of AI, Maike says. Modelling AI on alternate data sources and comparing them with results from the original dataset will be useful in uncovering and tackling biases, she explains.

Meharry Medical College in Nashville aims to use data to improve patient care and population health. It partnered with healthcare data company Clearsense to build a platform on Cloudera’s open-source and secure cloud architecture.

Researchers now have a better picture of population health and can break down results by race, social determinants of health, and other factors. Treatment becomes more specific and effective, without promoting bias.

This has also shaved three to five years off research, and has enabled researchers to build predictive models to tackle complex health problems like hypertension and diabetes.

Managing sensitive data

One challenge facing hospitals is data lineage. Patient data is sensitive, so healthcare providers want to know where data has come from, who has used it, and who it was shared with. This includes data transferred between hospitals, Maike states.

Data used in AI models must also be anonymised to avoid any breaches in privacy. IHiS has built an analytics platform that anonymises patient data prior to further analysis. That ensures confidential data is kept private, and enables analysts to request for information from disparate data sources.

Cloudera works with healthcare providers to track where data comes from and how it has been used, Maike says. Its tool can work with “all types of data”, whether it’s stored on the cloud, on-premise, or across a hybrid environment.

Its systems also use machine learning and analytics to detect anomalies and proactively secure healthcare data. This ensures only authorised parties have access to data and prevents data leaks.

AI holds massive potential for fighting Covid-19, but also in unlocking new frontiers in precision medicine. Organisations will need accurate, secure data in the journey ahead.