Using AI offers ways to improve Fabry diagnosis, care: Study

Artificial intelligence tech may reduce misdiagnoses, per scientists

Lindsey Shapiro, PhD avatar

by Lindsey Shapiro, PhD |

Share this article:

Share article via email
Various screens of computers running computations are shown.

Technologies using artificial intelligence (AI) have the ability to facilitate earlier diagnoses and better treatment for people living with rare conditions such as Fabry disease, according to a new review study by researchers in Europe.

In the study, the scientists discussed several ways AI has been applied in recent years to improve Fabry disease care.

Such approaches, the team noted, are also relevant for many other rare diseases, defined in the U.S. as those affecting fewer than 200,000 people, and in the European Union as having a prevalence of 1 in every 2,000 people.

“Patient diagnostic journeys may benefit from AI as these technologies may reduce the rate of misdiagnosis and shorten the period spent without appropriate medical care, thus lessening the psychological and physiological impact of disease,” the researchers wrote. The team noted that “the large amounts of data available from [electronic health records], medical imaging (standardized images) and DNA sequencing allow for large-population and patient-centred approaches.”

Their study, “Applying artificial intelligence to rare diseases: a literature review highlighting lessons from Fabry disease,” was published in the Orphanet Journal of Rare Diseases.

Recommended Reading
An illustration of a DNA strand highlights its double-helix structure.

‘Sequencing First’ Strategy Aims to Speed Diagnosis

Using AI may help ID optimal treatment strategies

While individually uncommon, rare diseases collectively affect as many as 400 million people globally, per some estimates. Among them is Fabry disease, a genetic disorder in which GLA gene mutations ultimately give rise to the harmful accumulation of certain fatty substances in cells. This leads to symptoms such as pain in the hands and feet and skin lesions, as well as more severe complications such as kidney failure and heart disease.

Despite the substantial burden rare conditions place on individuals and the healthcare system, these diseases are often unrecognized by physicians, causing patients to experience substantial diagnostic delays and misdiagnoses that impede their ability to receive the right treatment.

“The longer the odyssey, the higher the risk of disease progression and health impairment, especially for the [approximately] 5% of rare diseases for which there is an effective and specific treatment,” the researchers wrote.

One example is Fabry, for which there are available treatments — ones that, if initiated promptly, can prevent or slow the progression of organ damage.

In recent years, it’s become recognized that using AI — a type of computer technology that mimics human intelligence to perform complex tasks — may help to improve rare disease care. Such technology may aid healthcare professionals in making early and accurate diagnoses, predicting disease progression, and identifying optimal treatment strategies, according to the team.

Various types of AI are capable of analyzing large amounts of data and recognizing patterns and relationships much more quickly and accurately than the human eye.

In their review study, the scientists discussed some of the ways AI is being applied in rare diseases, focusing particularly on 20 studies related to Fabry disease as an example.

The team pointed out a few main ways AI has been used in Fabry research — including for screening electronic medical records to help identify undiagnosed Fabry patients who may have been overlooked.

The technology also has been used for analyzing large amounts of genetic sequencing data to detect gene mutations and determine if they’re likely to cause disease.

Further, the researchers noted, AI-based tools can learn to recognize specific features of a rare disease in clinical tests and distinguish it from other conditions. Such tools may help clinicians to hone in on a diagnosis — and also aid in monitoring disease progression, determining prognosis, and making treatment decisions.

Recommended Reading
Three people look at a tablet illustration.

Digital app for patient monitoring found helpful in small study

New technologies may help in diagnosing other rare diseases

In Fabry, such approaches can be used to identify disease symptoms, such as characteristic facial features, and signs of heart disease and brain changes.

Looking forward, the scientists see several opportunities for using AI in the rare disease space to improve diagnosis and monitoring. These approaches will likely work best when the AI model evaluates and integrates several different disease features, per the team.

Ultimately, the use of these technologies could shorten the amount of time physicians need to look at clinical data, allowing them more time to focus on patient care. It could also make it easier for clinicians without expertise in rare diseases to participate in the diagnostic process and make appropriate referrals.

Still, the researchers emphasized that “results generated by AI are not final.” It’s still critical that any findings are “validated by experts who will make decisions for optimizing patient management based on consideration of the patient ‘as a whole,’” the team wrote.

We are still at the early stages of the AI-based technological revolution that is transforming healthcare for both patients and professionals. … It is of utmost importance that we approach this transformation with an open mind [toward] innovation.

Beyond diagnosis and treatment, there are other uses for AI technology, the researchers noted — for example, incorporating it into wearable sensors for remote monitoring.

Using AI is not without challenges and limitations, however. For one, such technology usually learns to make its predictions by learning from large amounts of data. This can be hindered in rare diseases by a scarcity of available data.

The team noted a need for ongoing efforts to test and apply AI in clinical practice, and to establish policies that address ethical concerns related to its use.

“We are still at the early stages of the AI-based technological revolution that is transforming healthcare for both patients and professionals,” the researchers wrote. “It is of utmost importance that we approach this transformation with an open mind [toward] innovation.”