User login
TOPLINE:
Researchers have developed a tool that uses artificial intelligence (AI) to identify acute otitis media in children based on otoscopic videos. It may improve diagnosis of ear infections in primary care settings, the developers said.
METHODOLOGY:
- The developers relied on otoscopic videos of the tympanic membrane captured on smartphones connected to scopes.
- Their analysis focused on 1151 videos from 635 children, most younger than 3 years old, who were seen for sick or well visits at outpatient clinics in Pennsylvania from 2018 to 2023.
- The tool was trained to differentiate between patients who did and did not have acute otitis media.
TAKEAWAY:
- Out of an original pool of 1561 videos, 410 were excluded due to obstruction by cerumen. In the remaining videos, experts identified acute otitis media in 305 videos (26.5%) and no acute otitis media in 846 videos (73.5%).
- The tool achieved a sensitivity of 93.8% and specificity of 93.5%, with bulging of the tympanic membrane being the most indicative feature of acute otitis media, present in 100% of diagnosed cases, according to the researchers.
- Feedback from 60 parents was largely positive, with 80% wanting the tool to be used during future visits.
IN PRACTICE:
Based on the diagnostic accuracy of clinicians in other studies, “The algorithm exhibited higher accuracy than pediatricians, primary care physicians, and advance practice clinicians and, accordingly, could reasonably be used in these settings to aid with decisions regarding treatment,” the authors of the study wrote. “More accurate diagnosis of [acute otitis media] may help reduce unnecessary prescriptions of antimicrobials in young children,” they added.
Studies directly comparing the performance of the tool vs clinicians are still needed, however, according to an editorial accompanying the journal article.
“While the data from this study show the model’s accuracy (94%) is superior to historical accuracy of clinicians in diagnosing acute otitis media (84% or less), these data come from different studies not using the same definition for accuracy,” wrote Hojjat Salmasian, MD, MPH, PhD, and Lisa Biggs, MD, with Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania. “If we assume the model is confirmed to be highly accurate and free from bias, this model could truly transform care for patients with suspected acute otitis media.”
SOURCE:
Alejandro Hoberman, MD, with the University of Pittsburgh Medical Center Children’s Hospital of Pittsburgh, Pittsburgh, Pennsylvania, was the corresponding author of the study. It was published online in JAMA Pediatrics .
LIMITATIONS:
The study used convenience sampling and did not include external validation of the tool. The researchers lacked information about participant demographics and the reason for their clinic visit.
DISCLOSURES:
Three authors of the study are listed as inventors on a patent for a tool to diagnose acute otitis media. Two authors with Dcipher Analytics disclosed fees from the University of Pittsburgh for their work on an application programming interface during the study. The research was supported by the Department of Pediatrics at the University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania.
This article was created using several editorial tools, including AI, as part of the process. Human editors reviewed this content before publication. A version of this article appeared on Medscape.com.
TOPLINE:
Researchers have developed a tool that uses artificial intelligence (AI) to identify acute otitis media in children based on otoscopic videos. It may improve diagnosis of ear infections in primary care settings, the developers said.
METHODOLOGY:
- The developers relied on otoscopic videos of the tympanic membrane captured on smartphones connected to scopes.
- Their analysis focused on 1151 videos from 635 children, most younger than 3 years old, who were seen for sick or well visits at outpatient clinics in Pennsylvania from 2018 to 2023.
- The tool was trained to differentiate between patients who did and did not have acute otitis media.
TAKEAWAY:
- Out of an original pool of 1561 videos, 410 were excluded due to obstruction by cerumen. In the remaining videos, experts identified acute otitis media in 305 videos (26.5%) and no acute otitis media in 846 videos (73.5%).
- The tool achieved a sensitivity of 93.8% and specificity of 93.5%, with bulging of the tympanic membrane being the most indicative feature of acute otitis media, present in 100% of diagnosed cases, according to the researchers.
- Feedback from 60 parents was largely positive, with 80% wanting the tool to be used during future visits.
IN PRACTICE:
Based on the diagnostic accuracy of clinicians in other studies, “The algorithm exhibited higher accuracy than pediatricians, primary care physicians, and advance practice clinicians and, accordingly, could reasonably be used in these settings to aid with decisions regarding treatment,” the authors of the study wrote. “More accurate diagnosis of [acute otitis media] may help reduce unnecessary prescriptions of antimicrobials in young children,” they added.
Studies directly comparing the performance of the tool vs clinicians are still needed, however, according to an editorial accompanying the journal article.
“While the data from this study show the model’s accuracy (94%) is superior to historical accuracy of clinicians in diagnosing acute otitis media (84% or less), these data come from different studies not using the same definition for accuracy,” wrote Hojjat Salmasian, MD, MPH, PhD, and Lisa Biggs, MD, with Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania. “If we assume the model is confirmed to be highly accurate and free from bias, this model could truly transform care for patients with suspected acute otitis media.”
SOURCE:
Alejandro Hoberman, MD, with the University of Pittsburgh Medical Center Children’s Hospital of Pittsburgh, Pittsburgh, Pennsylvania, was the corresponding author of the study. It was published online in JAMA Pediatrics .
LIMITATIONS:
The study used convenience sampling and did not include external validation of the tool. The researchers lacked information about participant demographics and the reason for their clinic visit.
DISCLOSURES:
Three authors of the study are listed as inventors on a patent for a tool to diagnose acute otitis media. Two authors with Dcipher Analytics disclosed fees from the University of Pittsburgh for their work on an application programming interface during the study. The research was supported by the Department of Pediatrics at the University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania.
This article was created using several editorial tools, including AI, as part of the process. Human editors reviewed this content before publication. A version of this article appeared on Medscape.com.
TOPLINE:
Researchers have developed a tool that uses artificial intelligence (AI) to identify acute otitis media in children based on otoscopic videos. It may improve diagnosis of ear infections in primary care settings, the developers said.
METHODOLOGY:
- The developers relied on otoscopic videos of the tympanic membrane captured on smartphones connected to scopes.
- Their analysis focused on 1151 videos from 635 children, most younger than 3 years old, who were seen for sick or well visits at outpatient clinics in Pennsylvania from 2018 to 2023.
- The tool was trained to differentiate between patients who did and did not have acute otitis media.
TAKEAWAY:
- Out of an original pool of 1561 videos, 410 were excluded due to obstruction by cerumen. In the remaining videos, experts identified acute otitis media in 305 videos (26.5%) and no acute otitis media in 846 videos (73.5%).
- The tool achieved a sensitivity of 93.8% and specificity of 93.5%, with bulging of the tympanic membrane being the most indicative feature of acute otitis media, present in 100% of diagnosed cases, according to the researchers.
- Feedback from 60 parents was largely positive, with 80% wanting the tool to be used during future visits.
IN PRACTICE:
Based on the diagnostic accuracy of clinicians in other studies, “The algorithm exhibited higher accuracy than pediatricians, primary care physicians, and advance practice clinicians and, accordingly, could reasonably be used in these settings to aid with decisions regarding treatment,” the authors of the study wrote. “More accurate diagnosis of [acute otitis media] may help reduce unnecessary prescriptions of antimicrobials in young children,” they added.
Studies directly comparing the performance of the tool vs clinicians are still needed, however, according to an editorial accompanying the journal article.
“While the data from this study show the model’s accuracy (94%) is superior to historical accuracy of clinicians in diagnosing acute otitis media (84% or less), these data come from different studies not using the same definition for accuracy,” wrote Hojjat Salmasian, MD, MPH, PhD, and Lisa Biggs, MD, with Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania. “If we assume the model is confirmed to be highly accurate and free from bias, this model could truly transform care for patients with suspected acute otitis media.”
SOURCE:
Alejandro Hoberman, MD, with the University of Pittsburgh Medical Center Children’s Hospital of Pittsburgh, Pittsburgh, Pennsylvania, was the corresponding author of the study. It was published online in JAMA Pediatrics .
LIMITATIONS:
The study used convenience sampling and did not include external validation of the tool. The researchers lacked information about participant demographics and the reason for their clinic visit.
DISCLOSURES:
Three authors of the study are listed as inventors on a patent for a tool to diagnose acute otitis media. Two authors with Dcipher Analytics disclosed fees from the University of Pittsburgh for their work on an application programming interface during the study. The research was supported by the Department of Pediatrics at the University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania.
This article was created using several editorial tools, including AI, as part of the process. Human editors reviewed this content before publication. A version of this article appeared on Medscape.com.