Connect with us


Breaking new ground in AI diagnostics

Avatar photo



Clinithink’s advanced AI diagnostics technology has enabled clinicians at Rady Children’s Institute for Genomic Medicine, a world leader in the field of genomics, to perform the deep phenotyping of a critically ill five-week-old baby with an unknown condition within minutes to help identify his rare genetic disorder.  

A five-week old baby with a potentially fatal undiagnosed disorder was admitted to Rady Children’s Institute for Genomic Medicine (RCIGM) earlier this year.  The child had been previously healthy but became acutely unwell two hours prior to admission to the Neonatal Intensive Care Unit (NICU).

Testing showed he had encephalopathy, but the underlying cause was unknown. As the child’s condition deteriorated with multiple seizures, whole genome sequencing (WGS) was performed to evaluate whether he had an underpinning genetic disease. Over 1,500 of these diseases can cause encephalopathy and seizures, and each minute lost before a diagnosis increases the likelihood of permanent neurological damage or even death.

In this case, WGS took between 5 and 6 hours and narrowed down the potential diagnosis to about 500 rare diseases. Meanwhile, the infant’s condition was worsening and the clinicians were still in the dark as to how they should treat the child.

The next step in the process is known as deep phenotyping. This is the painstaking analysis of an individual patient’s characteristics and clinical findings which are then compared with over 12,000 phenotypes that are known to be associated with over 7000 rare diseases. The analysis complements genomic data and is essential to confirming a diagnosis. Conventionally, deep phenotyping is done manually and can take a day or longer, even when undertaken by highly trained geneticists.

In this case, the team at RCIGM were able to use a novel AI technology developed by a UK-based startup, Clinithink, which reduced the analysis time from ‘hours to minutes’.

“The challenge [with rare disease] is diagnosis. They can present in the first four weeks of life as a complete and total surprise,” Chris Tackaberry, CEO and co-founder at Clinithink, told Health Tech World.

“There is a normal pregnancy, normal delivery, nothing that would suggest there is going to be a problem. The child gets discharged from the hospital and the mother and baby are doing fine, and then three weeks later, bang, it becomes terribly sick within hours and is in the neonatal intensive care unit.

“All the obvious things get excluded very quickly. And then the physicians on the frontline are staring at a very scary prospect of trying to work out which of the 7000 known rare diseases this [child] has.”

Clinithink has developed innovative software systems to enable hospitals to review millions of clinical documents at a rapid pace. Its patented CLiX technology harnesses natural language processing.

Clinithink claims that CLiX is the world’s first AI system that is capable of ‘truly understanding’ unstructured medical notes.

The software was originally developed to save healthcare institutions time by automatically ‘interrogating’ documents such as discharge summaries, outpatient letters and referrals. But then, several years ago, Clinithink was introduced to the rare disease space by Alexion; a global biopharmaceutical company that selected Clinithink’s technology to support their strategic collaboration with RCIGM to accelerate the diagnosis of critically-ill newborns with rare genetic disorders.

Tackaberry was initially sceptical about whether it would work, but two years later, the company had developed a specific configuration for rare diseases called ‘CLiX focus’ which automates “deep phenotyping” to support diagnosis. The first proven use of the technology was in 2019.

The team at RCIGM used CLiX focus to complete the deep phenotyping step in just one minute. This helped to achieve a record end-to-end time of 13.5 hours from blood sample to diagnosis. The disease identified by whole gene sequencing, thiamine metabolism dysfunction syndrome 2 (THMD2), can be effectively treated with two vitamin supplements if diagnosed early.

The RCIGM published a letter in the New England Journal of Medicine (NEJM) describing how accelerated diagnosis powered by the first-of-its-kind AI software allowed the child to be put on treatment and be discharged from hospital three days after being admitted. Now at seven months of age the child is thriving.

Tragically, this was not the case for the child’s older sibling, who ten years earlier had been admitted to hospital with what is now believed to be the same condition. The child’s condition could not be diagnosed and died at the age of eleven months.

Although each rare disease affects as little as one in a million people, the total number of people with a form of rare disease is significantly higher. In the US, it is estimated that around 30% of the population have a rare disease, many of which go undiagnosed.

Another paper from RCIGM published in the American Journal of Human Genetics illustrates the potential of AI software to save lives and vastly reduce costs. The RCIGM acted as a hub for five of California’s neonatal intensive care units and over the course of the study had 184 blood samples from babies that potentially had a rare disease.

The medical records were processed by Clinithink’s solution which identified rare diseases in approximately one third of the patients. An economic analysis demonstrated that there was an average of $6,000 of savings per case.

CLiX focus is used by just a handful of institutions and is still at the “bleeding edge”, Tackaberry said. He is hesitant to predict when the technology will become more widely used, but, as with many new technologies that are introduced to the healthcare space, adoption can be slow.

“Building evidence like this is what’s going to be required to move the culture to a point where [clinicians] say ‘I can’t not do this’ […] and that’s probably one to five years away.

“That’s how technology gets adopted in healthcare. You get to the point where your professional obligation is to use a piece of technology which might have been rather novel but is now proven beyond doubt. It becomes a routine part of practice.

“That’s not an easy journey, but that’s the one that we’re on and we’re well down the track.”

Now that the technique is beginning to be proven, CLiX focus is being adopted by other organisations in the US and UK. The University of Utah is now routinely screening every patient that is admitted to the neonatal intensive care unit using CLiX focus.

Meanwhile, in the UK, Clinithink is halfway through a project with Barts Health NHS Trust which is using Clinithink’s technology to identify patients that are at risk of diabetic foot disease, a life threatening complication of diabetes. The company hopes the results from the project will be a springboard towards wider adoption across the NHS.

More stories on AI diagnostics

Are we on the verge of a quantum leap in taking diagnostics to patients 

AI boost to leukemia diagnosis 

AI to advance bowel cancer tests

Continue Reading
Click to comment

Leave a Reply

Your email address will not be published. Required fields are marked *

Trending stories