AI Algorithm Could Streamline Diagnoses for Children with Rare Diseases
Diagnosing rare diseases can be a lengthy and complex process, often leading to frustrating delays for families seeking answers about their child’s health.
A new artificial intelligence (AI) tool called Phen2Test shows promise in streamlining this process for pediatricians by helping them select the most appropriate genetic tests for children suspected of having rare genetic disorders.
Developed by researchers at Columbia University Vagelos College of Physicians & Surgeons, the algorithm mimics the decision-making process of experienced geneticists. By analyzing a patient’s medical record, the AI can recommend whether a comprehensive genetic test, such as whole exome sequencing (WES), should be ordered directly or if a more focused gene panel is more appropriate.
This cutting-edge technology could be a game-changer for diagnosing rare pediatric disorders, especially for pediatricians who may not have specialized training in genetics.
“Patients with rare diseases often experience prolonged diagnostic delays. Ordering appropriate genetic tests is crucial, yet challenging, especially for general pediatricians without genetic expertise,” the researchers noted. “This AI could help bridge that gap and expedite timely diagnosis and appropriate treatment.”
Traditional approaches often involve a step-wise process, starting with narrower gene panels specific to a suspected disease. If those tests are inconclusive,
doctors then move on to broader genetic analyses like WES or whole genome sequencing (WGS).
The Phen2Test, however, was trained on over 1,000
electronically stored health records of children with known rare diseases seen by geneticists at Columbia University Irving Medical Center. It learned to recognize patterns in a patient’s medical history, symptoms, and demographics, allowing it to predict which tests a specialist would likely order.
The AI’s performance was remarkably similar to that of experienced genetic specialists and significantly outperformed the accuracy of standard practice by general pediatricians who don’t regularly order genetic tests.
Beyond its accuracy, the AI fostered a model for cost-effectiveness
by estimating the Phen2Test could save around $536 per patient compared to using the traditional tiered approach. It could also be more cost-effective than jumping straight to
broad genetic analysis, potentially saving approximately $236 per patient.
While these findings look promising, the researchers emphasized that further validation
is needed before it can be implemented in clinical settings.
They also acknowledged that the model’s accuracy might vary in different healthcare settings, potentially needing adjustments for diverse resource allocation and clinical practices.
Nevertheless, this groundbreaking AI could hold key
to simplifying the path to diagnose and treat children with rare diseases, bringing much-needed clarity and hope to families confronting these often complex medical conditions.
How can Phen2Test streamline the diagnosis process for rare diseases in children?
## AI Algorithm Could Streamline Diagnoses for Children with Rare Diseases
**Interviewer:** Joining us today is Dr. [Guest Name], a [Guest Title] with expertise in [Guest Area of Expertise]. Dr. [Guest Name], thank you for being here.
**Dr. [Guest Name]:** Thank you for having me.
**Interviewer:** Let’s talk about this new AI tool making waves in the medical community, Phen2Test. How could this technology revolutionize the way we diagnose rare diseases in children?
**Dr. [Guest Name]:** This is incredibly promising. As you mentioned, diagnosing rare diseases can be a long and arduous process for families. Phen2Test uses AI to analyze a child’s medical record and essentially mimic the decision-making process of a seasoned geneticist. This helps pediatricians, who might not have specialized training in genetics, select the most appropriate genetic test right away. Instead of ordering a broad test that might not be necessary, the AI can recommend a focused gene panel or even whole exome sequencing if needed.
**Interviewer:** That sounds like it could save valuable time and resources.
**Dr. [Guest Name]:** Absolutely. Time is crucial for children with rare diseases. Early diagnosis leads to earlier treatment interventions and better outcomes. This AI has the potential to significantly reduce diagnostic delays [[1](https://www.uclahealth.org/news/article/ai-powered-tool-rare-diseases)].
**Interviewer:** So, you see this being adopted widely by pediatricians?
**Dr. [Guest Name]:** I certainly hope so. It’s a user-friendly tool designed to work seamlessly with existing electronic health record systems. By simplifying the testing process and leveraging the power of AI, Phen2Test can empower pediatricians to provide faster and more accurate diagnoses for their young patients.
**Interviewer:** This is truly groundbreaking. Thank you so much for sharing your insights, Dr. [Guest Name].
**Dr. [Guest Name]: **You’re welcome. It’s my pleasure.
