Correction: Feasibility of Multimodal Artificial Intelligence Using GPT-4 Vision for the Classification of Middle Ear Disease: Qualitative Study and Validation.

[This corrects the article DOI: 10.2196/58342.].

Feasibility of Multimodal Artificial Intelligence Using GPT-4 Vision for the Classification of Middle Ear Disease: Qualitative Study and Validation.

BACKGROUND: The integration of artificial intelligence (AI), particularly deep learning models, has transformed the landscape of medical technology, especially in the field of diagnosis using imaging and physiological data. In otolaryngology, AI has shown promise in image classification for middle ear diseases. However, existing models often lack patient-specific data and clinical context, limiting their universal applicability. The emergence of GPT-4 Vision (GPT-4V) has enabled a multimodal diagnostic approach, integrating language processing with image analysis. OBJECTIVE: In this study, we investigated the effectiveness of GPT-4V in diagnosing middle ear diseases by integrating patient-specific data with otoscopic images of the tympanic membrane. METHODS: The design of this study was divided into two phases: (1) establishing a model with appropriate prompts and (2) validating the ability of the optimal prompt model to classify images. In total, 305 otoscopic images of 4 middle ear diseases (acute otitis media, middle ear cholesteatoma, chronic otitis media, and otitis media with effusion) were obtained from patients who visited Shinshu University or Jichi Medical University between April 2010 and December 2023. The optimized GPT-4V settings were established using prompts and patients' data, and the model created with the optimal prompt was used to verify the diagnostic accuracy of GPT-4V on 190 images. To compare the diagnostic accuracy of GPT-4V with that of physicians, 30 clinicians completed a web-based questionnaire consisting of 190 images. RESULTS: The multimodal AI approach achieved an accuracy of 82.1%, which is superior to that of certified pediatricians at 70.6%, but trailing behind that of otolaryngologists at more than 95%. The model's disease-specific accuracy rates were 89.2% for acute otitis media, 76.5% for chronic otitis media, 79.3% for middle ear cholesteatoma, and 85.7% for otitis media with effusion, which highlights the need for disease-specific optimization. Comparisons with physicians revealed promising results, suggesting the potential of GPT-4V to augment clinical decision-making. CONCLUSIONS: Despite its advantages, challenges such as data privacy and ethical considerations must be addressed. Overall, this study underscores the potential of multimodal AI for enhancing diagnostic accuracy and improving patient care in otolaryngology. Further research is warranted to optimize and validate this approach in diverse clinical settings.

Epidemiology, aetiology and diagnosis of congenital hearing loss via hearing screening of 153 913 newborns.

BACKGROUND: Congenital hearing loss (HL), one of the most common paediatric chronic conditions, significantly affects speech and language development. Its early diagnosis and medical intervention can be achieved via newborn hearing screening. However, data on the prevalence and aetiology of congenital HL in infants who fail newborn hearing screening are limited. METHODS: The sample population included 153 913 infants who underwent newborn hearing screening, and the prevalence of congenital HL, defined as moderate to profound bilateral HL (BHL) or unilateral HL (UHL) (≥40 dB HL), in one prefecture of Japan was measured to minimize the loss-to-follow-up rate, a common factor affecting the screening procedure. Comprehensive aetiological investigation, including physiology, imaging, genetic tests, and congenital cytomegalovirus screening, was performed on children diagnosed with congenital HL. RESULTS: The calculated prevalence of congenital HL was 1.62 per 1000 newborns (bilateral, 0.84; unilateral, 0.77). More than half of the cases with congenital bilateral or severe to profound UHL showed genetic aetiology or cochlear nerve deficiency (CND), respectively. Approximately 4% and 6% of the cases of congenital BHL and UHL were associated with congenital cytomegalovirus infection and auditory neuropathy spectrum disorder, respectively. CONCLUSIONS: This is an epidemiological and comprehensive aetiological study of congenital HL, as determined via newborn hearing screening according to its severity and laterality, in a large-scale general population of a developed country. Our findings can serve as a reference for optimizing care and intervention options for children with HL and their families.

Algal growth inhibition by river water pollutants in the agricultural area around Lake Biwa, Japan.

An ecotoxicological study of river water discharged from the agricultural area around Lake Biwa was performed by using algal bioassays to guide chemical analysis. Water samples were collected once a week, at least, for 1 year starting in April 1997 and continuing until April 1998. The toxicities of the dissolved and particulate-adsorbed extracts of water samples were evaluated by the algal growth inhibition test and concentrations of individual pesticides were determined. Most of the river water that was collected during the periods when pesticides were applied to the paddy fields caused algal growth inhibition. Some extracts were found to contain herbicides (molinate, mefenacet, simetryn, or esprocarb) as major compounds. According to chemical assay and bioassay, simetryn was identified as the most toxic compound that caused algal growth inhibition.

Concentration and loading of pesticide residues in Lake Biwa basin (Japan).

The concentrations and loading rates of pesticides used in paddy fields were investigated over a period of 5 years in the Seta River, which is the only natural outlet of Lake Biwa. The lake's water catchment area is 3,174 km2, 20% of which contains paddy fields. Water samples were also collected in six rivers flowing into the lake in order to compare the contamination level and concentration profile. The pesticides analyzed were four herbicides (molinate, simetryn, oxadiazon, and thiobencarb), one fungicide (isoprothiolane), and two insecticides (diazinon and fenitrothion). Molinate, simetryn, oxadiazon and isoprothiolane were found at the higher frequencies with maximum concentrations of 1.1, 0.4, 0.1 and 0.5 microg,/l in the effluent river, one or two order of magnitude higher than that of effluent in influent rivers. These peak concentrations were observed during the application period in influent rivers and two or three weeks after that in effluent river. The frequency of occurrence of thiobencarb, diazinon, and fenitrothion was relatively low and their maximum concentrations in the effluent remained below 0.1 microg/l. The decrease of molinate, simetryn and oxadiazon concentrations in the effluent river were approximated by two straight lines plotted on semilogarithmic scale. Increased loading was induced by intense rainfall, which took place during the application period. Simetryn and isoprothiolane persisted in relatively high concentrations through the year were also influenced on its loading by the heavy rainfall in the following months. The percentages of the total amount of pesticides released through Lake Biwa to the basin in downstream were estimated to be 1.3-2.9% for molinate, 5.4-10.0% for simetryn, 0.6-1.3% for oxadiazon, 0.2-0.9% for thiobencarb, 1.8-6.6% for isoprothiolane, 0.3-2.1% for diazinon. and 0% for fenitrothion.