Japan
March 1, 2022
Gx Sweat Patch®
March 1, 2022Pediatrics
AI Finds Potential Treatment for Incurable Pediatric Brain Cancer
The diffuse intrinsic pontine glioma (DIPG) is a fatal infiltrating type of glioma with an overall survival time ranging from 9 months to a year. The poor prognosis of DIPG is partly due to its biological characteristics, mainly a very high prevalence of lysine to methionine substitutions that result in ACVR1 mutation in around 25% of cases. Currently, no therapy is efficient, and only radiotherapy elicits a therapeutic response; however, AI has proven helpful in accelerating the discovery of treatments for DIPG through drug repurposing.
Artificial intelligence algorithms created by the Benevolent platform build a biomedical knowledge graph, which illustrates known scientific relationships from literature and data to represent diseases biology. The graph comprises 1.2 billion associations, including gene-disease, gene-drug, drug-disease, and gene-gene connections. The knowledge graph extracts data from heterogeneous sources, including literature evidence, differential expression analysis, or clinical trial data, and connects entities through relationships (e.g., therapeutics or biological associations). AI algorithms help conclude the potential points of therapeutic intervention based on the interpretation of the knowledge graph. As a result, the combination of vandetanib and everolimus was identified as a possible therapeutic approach for DIPG.(1)
The diffuse intrinsic pontine glioma (DIPG) is a fatal infiltrating type of glioma with an overall survival time ranging from 9 months to a year. The poor prognosis of DIPG is partly due to its biological characteristics, mainly a very high prevalence of lysine to methionine substitutions that result in ACVR1 mutation in around 25% of cases.
Currently, no therapy is efficient, and only radiotherapy elicits a therapeutic response; however, AI has proven helpful in accelerating the discovery of treatments for DIPG through drug repurposing. Artificial intelligence algorithms created by the Benevolent platform build a biomedical knowledge graph, which illustrates known scientific relationships from literature and data to represent diseases biology. The graph comprises 1.2 billion associations, including gene-disease, gene-drug, drug-disease, and gene-gene connections. The knowledge graph extracts data from heterogeneous sources, including literature evidence, differential expression analysis, or clinical trial data, and connects entities through relationships (e.g., therapeutics or biological associations). AI algorithms help conclude the potential points of therapeutic intervention based on the interpretation of the knowledge graph. As a result, the combination of vandetanib and everolimus was identified as a possible therapeutic approach for DIPG.(1)
AI model can help diagnose pediatric buckle fractures
Artificial Intelligence (AI) algorithms can help diagnose difficult fractures, especially buckle types, in pediatric patients. A group of researchers, led by Dr. John Zech from Columbia University in New York City, created and tested a deep-learning model to screen for wrist fractures in different imaging tests. The trained algorithm interpreted a radiograph as positive if it identified at least one region with an 80% fracture probability. The algorithm was highly sensitive and specific on its own, and, in turn, residents used it to improve their accuracy in identifying fractures. Access to algorithm predictions significantly improved overall average resident accuracy in diagnosing fractures from 80% to 93%. Future studies aim to validate this model on an external data set and demonstrate the algorithm’s performance in real-life cases using prospective clinical trials.(2)
Artificial Intelligence (AI) algorithms can help diagnose difficult fractures, especially buckle types, in pediatric patients. A group of researchers, led by Dr. John Zech from Columbia University in New York City, created and tested a deep-learning model to screen for wrist fractures in different imaging tests. The trained algorithm interpreted a radiograph as positive if it identified at least one region with an 80% fracture probability.
The algorithm was highly sensitive and specific on its own, and, in turn, residents used it to improve their accuracy in identifying fractures. Access to algorithm predictions significantly improved overall average resident accuracy in diagnosing fractures from 80% to 93%. Future studies aim to validate this model on an external data set and demonstrate the algorithm’s performance in real-life cases using prospective clinical trials.(2)
Using Artificial Intelligence to Diagnose Rare Pediatric Disorders
A collaborative research study between the University of Utah Health in Salt Lake City, Utah, and Fabric Genomics Biotechnology company in Oakland, California, created a new artificial intelligence technology to diagnose rare disorders in critically ill children. Children with rare genetic diseases are often admitted to the NICU as soon as they are born and need prompt identification of the underlying conditions (likely congenital) for proper management and treatment. In these scenarios, it is of utmost importance to rapidly and accurately identify any DNA errors, a worthy task for the new GEM algorithm. The GEM model uses artificial intelligence and deep learning to find DNA errors leading to disease.
GEM allows whole-genome sequencing analysis by cross-referencing large databases from diverse populations, clinical disease information, and other scientific data repositories. A study tested GEM on whole-genome sequencing from 180 diagnosed pediatric cases from Rady’s Children’s San Diego. Outperforming competing tools, GEM identified the causative gene in its top two candidates 92% of the time and within a suitable time frame for these critical diseases. GEM can also detect structural variants that may cause up to 20% of genetic diseases, yet missed by the existing technology.(3)
A collaborative research study between the University of Utah Health in Salt Lake City, Utah, and Fabric Genomics Biotechnology company in Oakland, California, created a new artificial intelligence technology to diagnose rare disorders in critically ill children.
Children with rare genetic diseases are often admitted to the NICU as soon as they are born and need prompt identification of the underlying conditions (likely congenital) for proper management and treatment. In these scenarios, it is of utmost importance to rapidly and accurately identify any DNA errors, a worthy task for the new GEM algorithm. The GEM model uses artificial intelligence and deep learning to find DNA errors leading to disease.GEM allows whole-genome sequencing analysis by cross-referencing large databases from diverse populations, clinical disease information, and other scientific data repositories. A study tested GEM on whole-genome sequencing from 180 diagnosed pediatric cases from Rady’s Children’s San Diego. Outperforming competing tools, GEM identified the causative gene in its top two candidates 92% of the time and within a suitable time frame for these critical diseases. GEM can also detect structural variants that may cause up to 20% of genetic diseases, yet missed by the existing technology.(3
Cardiologs' AI Receives Clearance for Pediatric Use
Recent guidelines recommend that all children should be screened for the risk of sudden cardiac arrest or sudden cardiac death at a minimum every three years. An increasing number of physicians have been combining clinical expertise with AI-assisted analysis to detect abnormal heart rhythms in the pediatric population, leading to earlier intervention and better outcomes. Cardiologs, a medical technology company focused on cardiac diagnostics using AI, announced that it received the Food and Drug Administration (FDA) approval to use its new AI-powered cardiac diagnostic platform for pediatric cardiology. The new authorization was granted based on an analysis of the company’s improved deep learning algorithm.
The new model was reinforced by more than 20 million EKG recordings and was used to estimate a multinational sample of 10,000 EKG readings from patients in various pediatric age groups. Results showed that the updated algorithm improved sensitivity by 14% overall across significant arrhythmias and reduced the number of false positives.(4)
Recent guidelines recommend that all children should be screened for the risk of sudden cardiac arrest or sudden cardiac death at a minimum every three years. An increasing number of physicians have been combining clinical expertise with AI-assisted analysis to detect abnormal heart rhythms in the pediatric population, leading to earlier intervention and better outcomes.
Cardiologs, a medical technology company focused on cardiac diagnostics using AI, announced that it received the Food and Drug Administration (FDA) approval to use its new AI-powered cardiac diagnostic platform for pediatric cardiology. The new authorization was granted based on an analysis of the company’s improved deep learning algorithm. The new model was reinforced by more than 20 million EKG recordings and was used to estimate a multinational sample of 10,000 EKG readings from patients in various pediatric age groups. Results showed that the updated algorithm improved sensitivity by 14% overall across significant arrhythmias and reduced the number of false positives.(4)
Israeli technological prowess with pediatric medicine
Eosinophilic esophagitis, a chronic immune disease caused by food and other allergies, is a severe condition affecting children. Its diagnosis usually depends on highly trained doctors’ microscopic analysis of an esophageal biopsy. This process is not only time-consuming but liable to variability in the results. In September, the Technion-Israel Institute of Technology in Haifa, Israel, and the Cincinnati Children’s Hospital Medical Center in the US partnered to revolutionize pediatric medicine. By operating Technion’s technological prowess, the pathology department can analyze microscopic slides to find eosinophils allowing speedy and accurate diagnosis of the condition.(5)
Eosinophilic esophagitis, a chronic immune disease caused by food and other allergies, is a severe condition affecting children. Its diagnosis usually depends on highly trained doctors’ microscopic analysis of an esophageal biopsy. This process is not only time-consuming but liable to variability in the results.
In September, the Technion-Israel Institute of Technology in Haifa, Israel, and the Cincinnati Children’s Hospital Medical Center in the US partnered to revolutionize pediatric medicine. By operating Technion’s technological prowess, the pathology department can analyze microscopic slides to find eosinophils allowing speedy and accurate diagnosis of the condition.(5)
AI, telehealth, and sensor-based technologies facilitate autism diagnosis
Pediatricians screen for autism spectrum disorder (ASD) in children aged 18 to 24 months during health maintenance routine. However, this neurodevelopmental condition often eludes diagnosis until a child is four years or older. A new artificial intelligence diagnostic system, Behavior Imaging Solutions (BIS) IT Services and IT Consulting, may facilitate the diagnosis by pediatricians. Ronald and Sharon Oberleitner developed the Naturalistic Observation Diagnostic Assessment (NODA), consisting of 2 technology components.
The first, called SmartCapture, is a smartphone application that allows parents to complete a developmental questionnaire and record and upload videos of their child. The second component is an AI-powered sensor to recognize data patterns and identify markers associated with ASD. Looking ahead, various smart device–based applications might soon be available to assist parents in socializing children with ASD.(6)
Pediatricians screen for autism spectrum disorder (ASD) in children aged 18 to 24 months during health maintenance routine. However, this neurodevelopmental condition often eludes diagnosis until a child is four years or older.
A new artificial intelligence diagnostic system, Behavior Imaging Solutions (BIS) IT Services and IT Consulting, may facilitate the diagnosis by pediatricians. Ronald and Sharon Oberleitner developed the Naturalistic Observation Diagnostic Assessment (NODA), consisting of 2 technology components. The first, called SmartCapture, is a smartphone application that allows parents to complete a developmental questionnaire and record and upload videos of their child. The second component is an AI-powered sensor to recognize data patterns and identify markers associated with ASD. Looking ahead, various smart device–based applications might soon be available to assist parents in socializing children with ASD.(6)
Machine Learning Tool Predicts Devastating Intestinal Disease in Premature Infants
Necrotizing enterocolitis is a life-threatening intestinal disease of neonates. It is characterized by sudden and continuous intestinal inflammation and tissue death. Researchers from Columbia Engineering, based in New York City, and the University of Pittsburgh, Pennsylvania, have created a sensitive and specific early warning system for predicting enterocolitis in premature infants. The researchers hypothesized that a deep machine learning model using clinical, imaging, demographic, and microbiological data from preterm infants could detect patients at high risk for enterocolitis. Their prototype confirmed their hypothesis. Such a model would allow early intervention to prevent severe complications. Once the platform is ready, they will conduct a clinical trial to validate their design’s predictions in a real-time neonatal ICU cohort.(7)
Necrotizing enterocolitis is a life-threatening intestinal disease of neonates. It is characterized by sudden and continuous intestinal inflammation and tissue death. Researchers from Columbia Engineering, based in New York City, and the University of Pittsburgh, Pennsylvania, have created a sensitive and specific early warning system for predicting enterocolitis in premature infants.
The researchers hypothesized that a deep machine learning model using clinical, imaging, demographic, and microbiological data from preterm infants could detect patients at high risk for enterocolitis. Their prototype confirmed their hypothesis. Such a model would allow early intervention to prevent severe complications. Once the platform is ready, they will conduct a clinical trial to validate their design’s predictions in a real-time neonatal ICU cohort.(7)
References