BackgroundMelioidosis, caused by Burkholderia pseudomallei, is a potentially fatal infection endemic to Southeast Asia and northern Australia. Disseminated disease with bacteremia and multifocal organ involvement carries high mortality, particularly among patients with diabetes mellitus (DM) and hazardous alcohol use. Symmetrical peripheral gangrene (SPG) is a rare complication that may result in permanent disability.Case presentationA 57-year-old man with poorly controlled type 2 DM, chronic alcohol use, and malnutrition presented with septic shock and acute hypoxemic respiratory failure. Imaging revealed cavitary pneumonia and abscesses involving the liver, kidney, and prostate. Two independently collected blood cultures and a prostatic abscess aspirate yielded B. pseudomallei, confirmed through a multistep microbiological workflow with final VITEK 2 identification. The patient required vasopressors, mechanical ventilation, continuous venovenous hemofiltration, intravenous meropenem, and image-guided drainage of the prostatic abscess. Distal cyanosis was present at admission and progressed to dry gangrene of multiple fingers and toes. The symmetric acral distribution and absence of major-vessel obstruction were clinically consistent with SPG. Recurrent hyperkalemia and renal dysfunction prevented continued trimethoprim-sulfamethoxazole therapy, necessitating six months of doxycycline plus amoxicillin-clavulanate. Secondary infection and osteomyelitis of the gangrenous digits required multiple amputations. At follow-up 11 months after completion of eradication therapy, there was no evidence of recurrent melioidosis.ConclusionSPG is a rare but devastating complication of disseminated melioidosis. Prompt antimicrobial therapy, intensive organ support, source control, and early recognition of peripheral ischemia are important. Even after survival and successful treatment, substantial long-term functional disability may persist.

IntroductionDengue is a global health threat, with severe cases causing significant complications. Cytokines have been proposed as potential indicators of disease severity; however, the short plasma half-life and pre-analytical instability of free-form cytokines limit their clinical applicability. Exosome-encapsulated cytokines are protected by a lipid bilayer and may provide a more stable and integrated representation of host immune responses.MethodsIn this cross-sectional study, we analyzed single time-point plasma samples collected from patients during clinical evaluation between July and December 2023, with most samples obtained during the acute phase (within 7 days post-symptom onset). Plasma exosomes were isolated from patients with mild dengue, dengue with warning signs (DFWS), severe dengue (SD), other febrile illnesses, and healthy controls (HCs). Exosome-associated cytokines were quantified using a multiplex panel assessing 15 cytokines.ResultsAcross the five study groups, significant differences in plasma exosome cytokine levels were observed for interleukin (IL)-1β, IL-6, IL-10, IL-12, interferon-γ (IFN-γ), and tumor necrosis factor (TNF)-α (all p < 0.05). Post-hoc analyses further demonstrated that IL-6, IL-10, and TNF-α levels were significantly higher in both the DFWS and SD groups compared with HCs (all p < 0.01). In a subsequent analysis comparing mild dengue with the combined DFWS/SD group, significantly higher exosomal levels of IL-1β, IL-5, IL-10, IL-12, IL-13, and TNF-α were observed in the DFWS/SD group (all p < 0.05). Receiver operating characteristic (ROC) analysis showed that IL-1β, IL-10, and TNF-α moderately discriminated mild from severe cases, with area under the curve (AUC) values of approximately 0.7. An “all-positive” panel (TNF-α, IL-10, and IL-1β) achieved 84% sensitivity and 67.5% specificity for identifying DFWS/SD.ConclusionThese findings suggest that combined exosomal cytokine profiling may aid in disease severity stratification. However, given the cross-sectional design and moderate discriminatory performance, larger prospective studies are needed to validate its clinical applicability.

BACKGROUND:Since 2022, the global outbreak of Mpox caused by the Mpox virus (MPXV) Clade IIb has underscored the need for timely and decentralized molecular diagnostics. Although quantitative real-time PCR (qPCR) remains the diagnostic gold standard, its infrastructure requirements may limit accessibility in resource-constrained settings. Insulated isothermal PCR (iiPCR) represents a potential field-deployable alternative; however, clinical validation data for MPXV remains limited. METHODS:We prospectively enrolled 24 qPCR-confirmed Mpox patients between 2023 and 2024. Multiple specimen types were tested using the POCKIT Central MPXV iiPCR system and reference qPCR. Analytical sensitivity was assessed using serial viral dilutions. Viral load dynamics were evaluated using cycle threshold (Ct) values across specimen types and days post-symptom onset. Phylogenetic characterization was performed using a four-gene Sanger sequencing approach. RESULTS:Both iiPCR and qPCR demonstrated identical analytical limits of detection at 10-1 PFU/mL and complete qualitative concordance in cultured samples. Lesion-derived specimens showed the highest detection rates (100%) and consistently lower Ct values, whereas non-lesion specimens exhibited lower and more variable positivity. No clear differences in viral load-based diagnostic performance were observed between people living with HIV receiving antiretroviral therapy (ART) and people without HIV. Phylogenetic analysis confirmed exclusive circulation of MPXV Clade IIb. CONCLUSIONS:Under the conditions evaluated, the POCKIT Central iiPCR system achieved diagnostic performance comparable to qPCR, particularly for lesion-derived specimens. In this ART-treated HIV-prevalent cohort with preserved immune function, HIV co-infection was not associated with differences in diagnostic performance, supporting the potential utility of iiPCR for decentralized Mpox testing.

Background: Multimodal large language models (MLLMs) show promise in medical imaging, but their performance is highly dependent on prompt engineering. This study systematically evaluates how different prompting strategies affect diagnostic accuracy in clinical laboratory image interpretation. Methods: We evaluated five MLLMs (ChatGPT-4o, Gemini 2.0 Flash, Claude 3.5 Sonnet, Grok-2, and Perplexity Pro (Claude 3.5 Sonnet)) using 177 proficiency testing images across three domains: blood smears (n = 78), urinalysis (n = 50), and parasitology (n = 49). Three prompting approaches were compared: (1) complex multi-choice prompts with 20 diagnostic options, (2) zero-shot open-ended prompts, and (3) two-step descriptive-reasoning prompts. Images were sourced from the Taiwan Society of Laboratory Medicine external quality assurance archives with expert consensus diagnoses. Results: Zero-shot prompting significantly outperformed complex multi-choice prompts across all models and domains (p < 0.001). With zero-shot prompts, Gemini achieved 78.5% overall accuracy (urinalysis: 92.0%; parasitology: 75.5%; blood smears: 64.1%), representing a 17% improvement over complex prompts. Two-step descriptive-reasoning prompts further improved blood smear accuracy by 8-12% for top-performing models, but showed minimal benefit in urinalysis and parasitology. The re-query mechanism ("please reconsider") improved urinalysis accuracy by 7.6% but had a negligible effect on blood smears and parasitology. Conclusions: Prompting strategy critically determines MLLM diagnostic performance. Zero-shot approaches with minimal constraints consistently outperform complex multi-choice formats. The remarkable performance of general-purpose models in structured domains like urinalysis (>90% accuracy) demonstrates the considerable progress of multimodal AI. However, complex morphological tasks like blood smear interpretation require either specialized prompting techniques or domain-specific fine-tuning. These findings provide evidence-based guidance for optimizing AI integration in clinical laboratories.

In the current era of coronavirus disease 2019 (COVID-19), we were interested in searching for medications other than the currently available antiviral drugs Paxlovid and Molnupiravir that cause minimal side effects and do not harm the human body. Honeysuckle extract (HSE) is a traditional Chinese medicine (TCM) that has been shown to exert antiviral effects in other studies. However, no studies have indicated whether HSE has an inhibitory effect on SARS-CoV-2. We prepared HSEs from dried honeysuckle flowers. We performed a cell viability assay, median tissue culture infection dose (TCID50) assay, and qRT‒PCR, and calculated the virus titers using the Reed-Muench method to evaluate the inhibitory effects of aqueous and alcohol HSEs on SARS‒CoV‒2 and explore the possible underlying mechanisms. In this study, post-treatment with HSE resulted in dose-dependent decreases in both the RNA levels and TCID50 of SARS-CoV-2 in Vero E6 cells; treatment with 50 μg/ml and 100 μg/ml alcohol HSEs achieved up to 95.323% and 92.587% inhibition, respectively. Moreover, pre-treatment with aqueous HSEs effectively reduced the RNA levels, and TCID50 of SARS-CoV-2 by up to 99.684%, and alcohol HSEs achieved up to 99.921% inhibition; both of these effects occurred in a dose-dependent manner. The results suggest that HSEs may have the potential to prevent SARS-CoV-2 infection.

Individuals with asymptomatic SARS-CoV-2 infection can unknowingly transmit the virus, yet identifying such infections in vaccinated populations remains challenging. We conducted a longitudinal study of 129 vaccine recipients immunized with various combinations of SARS-CoV-2 spike (S) protein vaccine platforms. Sera were collected before the first dose (v1), at 2 weeks (v7) and 6 months (v8) after the third dose. Taiwan’s first major COVID-19 outbreak occurred between v7 and v8. We measured anti-nucleocapsid (anti-N) and anti-S IgG antibody titers by ELISA and assessed virus-neutralizing activity using live virus and pseudovirus assays. By developing an iterative serial screening method, we identified asymptomatic infections among unconfirmed cases. Our v7-v8 paired cohort resolved into three distinct groups: confirmed cases (21%), asymptomatic infections (17%), and uninfected cases (62%). In normalized v8 sera, confirmed cases exhibited an anti-S+++/anti-N+++ phenotype, while uninfected cases showed an anti-S+/anti-N+ phenotype. Statistical analysis validated a distinct asymptomatic group characterized by intermediate anti-S but baseline anti-N antibody levels (anti-S++/anti-N+). This approach enables more accurate estimates of infection prevalence and vaccine efficacy.

Background Rapid molecular diagnostics are critical for timely dengue virus (DENV) detection, especially in resource-limited settings. Reverse transcription-insulated isothermal polymerase chain reaction (RT-iiPCR) enables sample-to-answer amplification without thermocyclers. While this platform was previously validated, its diagnostic performance against contemporary viral genotypes and, critically, across different host immune statuses (primary vs. secondary infection) remains largely undefined. We assessed a DENV iiPCR assay against quantitative real-time reverse transcription PCR (qRT-PCR) during the 2023 outbreak in southern Taiwan. Methods : Acute-phase sera (n = 185) from febrile patients were tested by qRT-PCR, pan-DENV iiPCR, IgM/IgG ELISAs, and serotype-specific RT-PCR/iiPCR. Diagnostic performance was evaluated with qRT-PCR as a reference. Subgroup analyses examined effects of days post-symptom onset (PSO), immune status (primary vs secondary infection), and serotype on iiPCR sensitivity. Results : Pan-iiPCR achieved 88.4 % sensitivity and 100 % specificity versus qRT-PCR. Sensitivity peaked within 3 days PSO (96.8 %) and in primary infections (95.0 %) but declined after day 3 (53.3 %) and in secondary infections (82.2 %). Combining pan-iiPCR with IgM testing raised sensitivity to ≥ 99 % beyond the viremic window. iiPCR correctly typed 83.5 % (71/85) of DENV-1 and 94.3 % (66/70) of DENV-2. Because pan-iiPCR served as the screening step, 18 RT-PCR–positive specimens were pan-iiPCR negative and therefore did not proceed to iiPCR serotyping. Conclusions : iiPCR enables rapid early-phase detection and is useful when paired with serology. However, reduced sensitivity in low-viremia and secondary infections, as well as incomplete serotyping, limit standalone use. Primer optimization and integration with IgM testing may enhance outbreak utility.

Background The continued evolution of SARS-CoV-2 Omicron sublineages requires ongoing genomic and epidemiological surveillance. This study investigated the genomic epidemiology, national trends, and clinical aspects of SARS-CoV-2 in Taiwan from November 2023 to February 2025. Methods We analyzed the molecular and clinical characteristics of a local cohort (n = 22) from southern Taiwan. Case and mortality trends were evaluated using nationwide surveillance data, considering changes in case definitions. Broader genomic surveillance included 1782 Taiwanese sequences from GISAID to characterize the distribution and evolution of Omicron subvariants, considering the rollout timeline of XBB.1.5 and JN.1 vaccines. Results Genomic surveillance revealed a major epidemiological shift driven by variant succession: JN.1 and its descendants (e.g., KP.2, KP.3) replaced XBB-related subvariants, coinciding with infection waves in early and mid-2024. A more stringent national case definition for “severe complicated COVID-19” implemented in September 2024, substantially altered epidemiological metrics, reducing reported cases but increasing case fatality. In the local cohort, no significant associations were observed between sex, age, Ct value, comorbidity number, or vaccination status and hospitalization or severity, likely due to limited sample size. Nationally, individuals aged ≥ 65 years were disproportionately represented among severe cases and deaths, with risk amplified in those with comorbidities and low recent XBB.1.5/JN.1 vaccination coverage. Conclusions Taiwan experienced dynamic evolution of SARS-CoV-2 Omicron, dominated by JN.1 and its derivatives, which drove successive infection waves. Changes in national case definitions created substantial surveillance artifacts complicating interpretation of disease burden. Although national data reaffirm advanced age as the key risk factor for severe outcomes, these findings underscore the importance of targeted vaccination strategies and integrated surveillance systems capable of distinguishing the effects of viral evolution from policy-driven reporting changes.

After the previous major dengue fever (DF) outbreaks in 2014 and 2015 in Taiwan, the second-largest DF outbreak re-emerged in 2023. A total of 178 patients with laboratory-confirmed dengue virus (DENV) infection, including 92 DENV-1 and 86 DENV-2 cases, were enrolled in this study conducted during the 2023 dengue outbreak in Kaohsiung City, Taiwan. This study aimed to analyze epidemiological characteristics, clinical severity, and risk factors for severe dengue (SD), as well as the diagnostic implications of the non-structural protein 1 (NS1) antigen rapid test. Patients infected with DENV-2 exhibited significantly older age, higher incidence of secondary infections, diabetes mellitus (DM), hypertension (HT), and longer hospital stays than patients infected with DENV-1. Multivariate analysis revealed that older age (age ≥65), secondary dengue infection, DM, and HT were significant independent predictors of SD. Compared with non-SD cases, SD patients were significantly more likely to be older (age ≥65), to exhibit a higher incidence of secondary infections and a greater prevalence of chronic diseases, including DM and HT. Notably, dengue-confirmed patients with negative NS1 results had a shorter duration since symptom onset (p < 0.001). Our DENV-1 and DENV-2 isolates are related to strains from neighboring Asian countries. Our findings emphasize the important factors of old age, secondary infections, and chronic diseases that contributed to dengue severity. We should meticulously manage these high-risk groups to prevent dengue progression. Screening incoming travelers for DF during the epidemic season will be an important measure to prevent the introduction of DENV into Taiwan.

BACKGROUND:Dengue virus (DENV) is a mosquito-borne flavivirus that remains a global public health threat, particularly in tropical regions where it is endemic. Accurate and timely diagnosis is critical for optimal patient care and effective outbreak containment. While reverse transcription-polymerase chain reaction (RT-PCR) remains the gold standard for dengue diagnosis, nonstructural protein 1 (NS1) rapid diagnostic tests (RDTs) offer a more accessible alternative. However, their performance may vary with infection type, serotype, and the timing of sample collection. METHODS:We evaluated the diagnostic performance of two NS1 RDTs, Abbott NS1 and PBF NS1, against RT-PCR as the reference standard. A total of 183 dengue-positive and 40 dengue-negative samples were obtained during the 2023 dengue outbreak in Kaohsiung, Taiwan. Sensitivity, specificity, and predictive values were comprehensively assessed based on infection type, serotype, and post-symptom onset (PSO) intervals. Additionally, we analyzed the diagnostic benefits of combining NS1 detection with anti-DENV IgM serology. RESULTS:Abbott NS1 exhibited notably higher sensitivity (87.4 % vs. 83.1 %) and specificity (100 % vs. 97.5 %) compared to PBF NS1, though the differences were not statistically significant. Sensitivity varied by PSO, with Abbott NS1 achieving significantly higher sensitivity beyond three days (97.4 % vs. 76.9 %, P = 0.034). In secondary infections, Abbott NS1 outperformed PBF NS1 (86.8 % vs. 79.2 %). When combined with anti-DENV IgM serology, both RDTs demonstrated improved diagnostic sensitivity, particularly in secondary infections. CONCLUSIONS:Our findings underscore the importance of context-specific diagnostic tools. Combining NS1 antigen detection with IgM serology enhances diagnostic accuracy. While Abbott NS1 demonstrated slightly better performance, PBF NS1 remains a viable and non-inferior alternative.

[This corrects the article DOI: 10.1371/journal.pone.0214328.].

BACKGROUND:Omicron, a severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) variant, entered Taiwan at the end of 2021. The Taiwanese government ended its "zero-COVID" policy in March 2022. Multiple coronavirus disease 2019 (COVID-19) outbreaks began in April 2022. We monitored the replacement of Omicron subvariants after BA.1/BA.2 and analyzed their correlation with COVID-19 outbreaks. METHODS:We collected SARS-CoV-2 real-time qRTPCR-positive nasopharyngeal swabs from Kaohsiung Medical University Hospital (KMUH), Kaohsiung City, Taiwan, and performed sequencing for specimens exhibiting a cytopathic effect in Vero E6 cells to determine their clades and lineages. We analyzed the medical records of COVID-19 patients and identified hospitalization risk factor(s). We retrieved SARS-CoV-2 sequences identified in Taiwan from GISAID and analyzed their correlation with COVID-19 data from the Taiwan Centers for Disease Control. RESULTS:We analyzed the phylogenesis of KMUH-47 to KMUH-104 (SARS-CoV-2 isolates identified herein) and all of the Omicron subvariants from BA.5 to XBB.1 (n = 1930). Age and comorbidities were hospitalization risk factors. Men generally exhibited a greater fatality rate than women. COVID-19-related deaths predominantly occurred in individuals over 70 years old. The COVID-19-related case fatality rate increased as nucleotide (NT) and amino acid (AA) substitutions increased. The number of COVID-19-related cases and deaths progressively decreased with each outbreak between August 2022 and October 2023. CONCLUSION:Hospitalization was associated with age and the presence of comorbidities. COVID-19-related fatality was linked to sex, age, and the accumulation of NT and AA substitutions in emerging Omicron subvariants.

Background Dengue virus (DENV) is the leading cause of arboviral diseases in humans worldwide. Currently Dengvaxia, the first dengue vaccine licensed in 20 countries, was recommended for DENV seropositive individuals aged 9–45 years. Studying dengue seroprevalence can improve our understanding of the epidemiology and transmission dynamics of DENV, and facilitate future intervention strategies and assessment of vaccine efficacy. Several DENV envelope protein-based serological tests including IgG and IgG-capture enzyme-linked immunosorbent assays (ELISAs) have been employed in seroprevalence studies. Previously DENV IgG-capture ELISA was reported to distinguish primary and secondary DENV infections during early convalescence, however, its performance over time and in seroprevalence study remains understudied. Methods In this study, we used well-documented neutralization test- or reverse-transcription-polymerase-chain reaction-confirmed serum/plasma samples including DENV-naïve, primary and secondary DENV, primary West Nile virus, primary Zika virus, and Zika with previous DENV infection panels to compare the performance of three ELISAs. Results The sensitivity of the InBios IgG ELISA was higher than that of InBios IgG-capture and SD IgG-capture ELISAs. The sensitivity of IgG-capture ELISAs was higher for secondary than primary DENV infection panel. Within the secondary DENV infection panel, the sensitivity of InBios IgG-capture ELISA decreased from 77.8% at < 6 months to 41.7% at 1–1.5 years, 28.6% at 2–15 years and 0% at > 20 years (p < 0.001, Cochran-Armitage test for trend), whereas that of IgG ELISA remains 100%. A similar trend was observed for SD IgG-capture ELISA. Conclusions Our findings demonstrate higher sensitivity of DENV IgG ELISA than IgG-capture ELISA in seroprevalence study and interpretation of DENV IgG-capture ELISA should take sampling time and primary or secondary DENV infection into consideration.

The Omicron variant of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), first detected in October 2021, possessed many mutations compared to previous variants. We aimed to identify and analyze SARS-CoV-2 Omicron subvariants among coronavirus disease 2019 (COVID-19) patients between January 2022 and September 2022 in Taiwan. The results revealed that BA.2.3.7, featuring K97E and G1251V in the spike protein compared with BA.2, emerged in March 2022 and persistently dominated between April 2022 and August 2022, resulting in the largest COVID-19 outbreak since 2020. The accumulation of amino acid (AA) variations, mainly AA substitution, in the spike protein was accompanied by increasing severity in Omicron-related COVID-19 between April 2022 and January 2023. Older patients were more likely to have severe COVID-19, and comorbidity was a risk factor for COVID-19-related mortality. The accumulated case fatality rate (CFR) dropped drastically after Omicron variants, mainly BA.2.3.7, entered Taiwan after April 2022, and the CFR was 0.16% in Taiwan, which was lower than that worldwide (0.31%) between April 2021 and January 2023. The relatively low CFR in Omicron-related COVID-19 patients can be attributed to adjustments to public health policies, promotion of vaccination programs, effective antiviral drugs, and the lower severity of the Omicron variant.

BackgroundDengue virus serotype 2 (DENV-2) was the major serotype in the 2015 dengue outbreak in Taiwan, while DENV-1 and DENV-3 were dominant between 2005 and 2014. We aimed to investigate whether DENV-2 contributed to disease severity and mortality in the outbreak in Kaohsiung city, Taiwan.MethodsWe collected serum samples from dengue patients to detect the presence of DENV and determine the serotypes by using quantitative reverse transcription-polymerase chain reaction. Our cohorts comprised 105 DENV-1-infected cases and 1,550 DENV-2-infected cases. Demographic data, DENV serotype, and comorbidities were covariates for univariate and multivariate analyses to explore the association with severity and mortality.ResultsThe results suggested that DENV-1 persisted and circulated, while DENV-2 was dominant during the dengue outbreak that occurred between September and December 2015. However, DENV-2 did not directly contribute to either severity or mortality. Aged patients and patients with diabetes mellitus (DM) or moderate to severe chronic kidney disease (CKD) had a higher risk of developing severe dengue. The mortality of dengue patients was related to a higher Charlson comorbidity index score and severe dengue. Among DENV-2-infected patients and older patients, preexisting anti-dengue IgG, DM, and moderate to severe CKD were associated with severe dengue. Moreover, female sex and severe dengue were associated with a significantly higher risk of death.ConclusionsOur findings highlight the importance of timely serological testing in elderly patients to identify potential secondary infections and focus on the meticulous management of elderly patients with DM or moderate to severe CKD to reduce dengue-related death.

In Taiwan, coronavirus disease 2019 (COVID-19) involving the delta variant occurred after that involving the alpha variant in 2021. In this study, we aimed to analyze the Delta variant. A total of 318 patients in Taiwan infected with delta variants were identified. The case fatality rate (CFR) of patients infected with delta variants was 0.94% in Taiwan compared with that of those infected with alpha variants (5.95%). The possible reasons for the low CFR might be hybrid immunity due to infection and rapid promotion of the COVID-19 vaccination program during the alpha variant outbreak. We identified three 21J delta variants. Two long gene deletions were detected in these severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) isolates: ORF7aΔ91 in KMUH-8 and SpikeΔ30 in KMUH-9. Protein structure prediction indicates that ORF7aΔ91 results in malfunction of NS7a as an interferon antagonist and that SpikeΔ30 results in a truncated spike protein (N679-A688del), resulting in a lower infection rate compared with the delta variant without these deletions. The impact of these two deletions on SARS-CoV-2-associated pathogenesis deserves further investigation. Delta variants still exist in many regions in the omicron era, and the backbone of the delta variant genome possibly spread worldwide in the form of delta-omicron hybrids (deltacron; e.g., XBC.1 and XAY.2), which casts a potential threat to public health. Our study further highlighted the importance of more understanding of the delta variants.

Images in children’s books convey messages about society and the roles that people play in it. Understanding these messages requires systematic measurement of who is represented. Computer vision face detection tools can provide such measurements; however, state-of-the-art face detection models were trained with photographs, and 80% of images in children’s books are illustrated; thus existing methods both misclassify and miss classifying many faces. In this paper, we introduce a new approach to analyze images using AI tools, resulting in data that can assess representation of race, gender, and age in both illustrations and photographs in children’s books. We make four primary contributions to the fields of deep learning and social sciences: (1) We curate an original face detection data set (IllusFace 1.0) by manually labeling 5,403 illustrated faces with bounding boxes. (2) We train two AutoML-based face detection models for illustrations: (i) using IllusFace 1.0 (FDAI); (ii) using iCartoon, a publicly available data set (FDAI iC), each optimized for illustrated images, detecting 2.5 times more faces in our testing data than the established face detector using Google Vision (FDGV). (3) We curate a data set of the race, gender, and age of 980 faces manually labeled by three different raters (CBFeatures 1.0). (4) We train an AutoML feature classification model (FCA) using CBFeatures 1.0. We compare FCA with the performance of another AutoML model that we trained on UTKFace, a public data set (FCA UTK) and of an established model using FairFace (FCF). Finally, we examine distributions of character identities over the last century across the models. We find that FCA is 34% more accurate than FCF in its race predictions. These contributions provide tools to educators, caregivers, and curriculum developers to assess the representation contained in children’s content.

[This corrects the article DOI: 10.3389/fmed.2022.869818.].