Assessing unknown potential-quality and limitations of different large language models in the field of otorhinolaryngology.

BACKGROUND: Large Language Models (LLMs) might offer a solution for the lack of trained health personnel, particularly in low- and middle-income countries. However, their strengths and weaknesses remain unclear. AIMS/OBJECTIVES: Here we benchmark different LLMs (Bard 2023.07.13, Claude 2, ChatGPT 4) against six consultants in otorhinolaryngology (ORL). MATERIAL AND METHODS: Case-based questions were extracted from literature and German state examinations. Answers from Bard 2023.07.13, Claude 2, ChatGPT 4, and six ORL consultants were rated blindly on a 6-point Likert-scale for medical adequacy, comprehensibility, coherence, and conciseness. Given answers were compared to validated answers and evaluated for hazards. A modified Turing test was performed and character counts were compared. RESULTS: LLMs answers ranked inferior to consultants in all categories. Yet, the difference between consultants and LLMs was marginal, with the clearest disparity in conciseness and the smallest in comprehensibility. Among LLMs Claude 2 was rated best in medical adequacy and conciseness. Consultants' answers matched the validated solution in 93% (228/246), ChatGPT 4 in 85% (35/41), Claude 2 in 78% (32/41), and Bard 2023.07.13 in 59% (24/41). Answers were rated as potentially hazardous in 10% (24/246) for ChatGPT 4, 14% (34/246) for Claude 2, 19% (46/264) for Bard 2023.07.13, and 6% (71/1230) for consultants. CONCLUSIONS AND SIGNIFICANCE: Despite consultants superior performance, LLMs show potential for clinical application in ORL. Future studies should assess their performance on larger scale.

Noise annoyance due to different sources is associated with tinnitus presence and distress in the general population.

BACKGROUND: The pathophysiology of tinnitus is not yet fully understood. Although there is a large amount of evidence associating traffic noise exposure with non-auditory health outcomes, there is no evidence regarding the impact of noise annoyance on auditory disorders such as tinnitus. OBJECTIVE: Thus, we aimed to investigate the association between noise annoyance due to different sources and tinnitus presence and distress in the general population. METHODS: Data of 6813 participants from a large German population-based cohort were used (Gutenberg Health Study). Participants were asked about the presence of tinnitus and how much they were bothered by it. In addition, information on annoyance from road traffic, aircraft, railways, industrial, and neighborhood noise during the day and sleep was collected through validated questionnaires. RESULTS: The prevalence of tinnitus was 27.3%, and the predominant sources of noise annoyance in these subjects were aircraft, neighborhood, and road traffic noise. Overall, logistic regression results demonstrated consistent positive associations between annoyance due to different noise sources and prevalent risk of tinnitus with increases in odds ratios ranging from 4 to 11% after adjustment for sex, age, and socioeconomic status. Likewise, consistent increases in odds ratios were observed for tinnitus distress in subjects with prevalent tinnitus. For instance, neighborhood noise annoyance during the sleep was associated with a 26% increase in tinnitus distress (OR 1.26, 95% CI 1.13; 1.39). IMPACT: This is the first study investigating the association between noise annoyance and tinnitus presence and distress in a large cohort of the general population. Our results indicate consistent and positive associations between various sources of noise annoyance and tinnitus. These unprecedented findings are highly relevant as noise annoyance and tinnitus are widespread. The precise etiology and locus of tinnitus remain unknown, but excessive noise exposure is thought to be among the major causes. This study suggests that transportation and neighborhood noise levels thought merely to contribute to annoyance and non-auditory health effects may be sufficient to cause or exacerbate tinnitus.

Age standardization and time-of-day performance for the Oldenburg Sentence Test (OLSA): results from the population-based Gutenberg Health Study.

PURPOSE: The Oldenburg Sentence Test (OLSA) is a German matrix test designed to determine speech recognition thresholds (SRT). It is widely used for hearing-aids and cochlear implant fitting, but an age-adjusted standard is still lacking. In addition, knowing that the ability to concentrate is an important factor in OLSA performance, we hypothesized that OLSA performance would depend on the time of day it was administered. The aim of this study was to propose an age standardization for the OLSA and to determine its diurnal performance. METHODS: The Gutenberg Health Study is an ongoing population-based study and designed as a single-centre observational, prospective cohort study. Participants were interviewed about common otologic symptoms and tested with pure-tone audiometry and OLSA. Two groups-subjects with and without hearing loss-were established. The OLSA was performed in two runs. The SRT was evaluated for each participant. Results were characterized by age in 5-year cohorts, gender and speech recognition threshold (SRT). A time stamp with an hourly interval was also implemented. RESULTS: The mean OLSA SRT was - 6.9 ± 1.0 dB (group 1 male) and - 7.1 ± 0.8 dB (group 1 female) showing an inverse relationship with age in the whole cohort, whereas a linear increase was observed in those without hearing loss. OLSA-SRT values increased more in males than in females with increasing age. No statistical significance was found for the diurnal performance. CONCLUSIONS: A study with 2900 evaluable Oldenburg Sentence Tests is a novelty and representative for the population of Mainz and its surroundings. We postulate an age- and gender-standardized scale for the evaluation of the OLSA. In fact, with an intergroup standard deviation (of about 1.5 dB) compared to the age dependence of 0.7 dB/10 years, this age normalization should be considered as clinically relevant.

ChatGPT Versus Consultants: Blinded Evaluation on Answering Otorhinolaryngology Case-Based Questions.

BACKGROUND: Large language models (LLMs), such as ChatGPT (Open AI), are increasingly used in medicine and supplement standard search engines as information sources. This leads to more "consultations" of LLMs about personal medical symptoms. OBJECTIVE: This study aims to evaluate ChatGPT's performance in answering clinical case-based questions in otorhinolaryngology (ORL) in comparison to ORL consultants' answers. METHODS: We used 41 case-based questions from established ORL study books and past German state examinations for doctors. The questions were answered by both ORL consultants and ChatGPT 3. ORL consultants rated all responses, except their own, on medical adequacy, conciseness, coherence, and comprehensibility using a 6-point Likert scale. They also identified (in a blinded setting) if the answer was created by an ORL consultant or ChatGPT. Additionally, the character count was compared. Due to the rapidly evolving pace of technology, a comparison between responses generated by ChatGPT 3 and ChatGPT 4 was included to give an insight into the evolving potential of LLMs. RESULTS: Ratings in all categories were significantly higher for ORL consultants (P<.001). Although inferior to the scores of the ORL consultants, ChatGPT's scores were relatively higher in semantic categories (conciseness, coherence, and comprehensibility) compared to medical adequacy. ORL consultants identified ChatGPT as the source correctly in 98.4% (121/123) of cases. ChatGPT's answers had a significantly higher character count compared to ORL consultants (P<.001). Comparison between responses generated by ChatGPT 3 and ChatGPT 4 showed a slight improvement in medical accuracy as well as a better coherence of the answers provided. Contrarily, neither the conciseness (P=.06) nor the comprehensibility (P=.08) improved significantly despite the significant increase in the mean amount of characters by 52.5% (n= (1470-964)/964; P<.001). CONCLUSIONS: While ChatGPT provided longer answers to medical problems, medical adequacy and conciseness were significantly lower compared to ORL consultants' answers. LLMs have potential as augmentative tools for medical care, but their "consultation" for medical problems carries a high risk of misinformation as their high semantic quality may mask contextual deficits.

Vertigo and its burden of disease-Results from a population-based cohort study.

Objectives: Vertigo describes symptoms of abnormal movement of the environment or the patient's own body. As such, it affects patients' quality of life, prevents them from following their daily activities, and increases healthcare utilization. The Global Burden of Disease Project aims to quantify morbidity and mortality worldwide. In 2013, a separate disability weight for vertigo was introduced. The aim of this study is to estimate the symptom burden of disease caused by vertigo. Methods: This study analyzes data from the Gutenberg Health Study (GHS). The GHS is a population-based cohort study representative of the city of Mainz and its district. Participants were asked whether they suffered from vertigo and, if so, how bothered they felt by it, rating their distress on a six-level scale from 1 = little stressful to 6 = extremely stressful. Results: Eight thousand five hundred and nineteen participants could be included in the study. The overall prevalence of vertigo was 21.6% (95%-confidence interval [CI] [20.7%; 22.5%]). Vertigo prevalence peaked in the age group of 55-64 years. Vertigo annoyance averaged 2.42 (±1.28). When an annoyance of 3-6 was considered bothersome, the prevalence of bothersome vertigo was 8.1 % (95%-CI [7.5%; 8.7%]). Age-standardized to the European Standard Population 2013, vertigo caused a burden of 2102 years lived with disability per 100,000 population. Conclusion: In this study, it was found that one in five people suffer at least occasionally from vertigo. This result suggests a significant burden of disease. This burden is reported at the symptom level. Future studies are needed to attribute the burden to specific causes. Level of Evidence: 2.

Anatomic Features of the Nasal and Pharyngeal Region Do Not Influence PAP Therapy Response.

The objective of this study was to investigate to which extent anatomic features of the nasal and pharyngeal region contribute to the severity of obstructive sleep apnea (OSA) and positive airway pressure (PAP) therapy response. Therefore, 93 patients (mean age 57.5 ± 13.0 years, mean body mass index 32.2 ± 5.80 kg/m2, 75 males, 18 females) diagnosed with OSA who subsequently started PAP therapy were randomly selected from the databank of a sleep laboratory of a tertiary university medical center. Patients were subdivided based on nasal anatomy (septal deviation, turbinate hyperplasia, their combination, or none of the above), pharyngeal anatomy (webbing, tonsillar hyperplasia, their combination, or none of the above), and (as a separate group) tongue base anatomy (no tongue base hyperplasia or tongue base hyperplasia). Then, polysomnographic data (e.g., arousal index, ARI; respiratory disturbance index, RDI; apnea index, AI; hypopnea index, HI; and oxygen desaturation index, ODI) of diagnostic polysomnography (PSG) and PAP therapy control PSG were collected, grouped, and evaluated. Septal deviation, turbinate hyperplasia, or their combination did not significantly affect the assessed PSG parameters or the response to PAP therapy compared with patients without nasal obstruction (p > 0.05 for all parameters). Accordingly, most PSG parameters and the response to PAP therapy were not significantly affected by webbing, tonsil hyperplasia, or their combination compared with patients without pharyngeal obstruction (p > 0.05 for RDI, AI, HI, and ODI, respectively). However, in the pharyngeal anatomy group, ARI was significantly higher in patients with tonsil hyperplasia (p = 0.018). Further, patients with tongue base hyperplasia showed a significantly higher HI in the diagnostic PSG (p = 0.025) compared with patients with normal tongue base anatomy, but tongue base anatomy did not significantly affect the response to PAP therapy (p > 0.05 for all parameters). The influence of anatomic features of the nasal and pharyngeal region on PAP therapy response appears to be small, and generalizability of these results requires further studies.

Drug-Induced Sleep Endoscopy Findings and Hypoglossal Nerve Stimulation Therapy Outcomes.

Hypoglossal-nerve stimulation (HGNS) is an established second-line therapy for patients with obstructive sleep apnea (OSA). Existing studies investigating the effect of preoperative drug-induced sleep endoscopic (DISE) findings on HGNS outcomes have mainly focused on the apnea/hypopnea index (AHI) among polysomnography (PSG) parameters, and have less frequently tested other PSG parameters such as the apnea index (AI), hypopnea index (HI), oxygen desaturation index (ODI), snoring index, and arousal index, or patient-reported excessive daytime sleepiness. The aim of this study was to investigate the correlation between DISE findings and the above-mentioned metrics after HGNS therapy. We only included patients with DISE findings providing detailed information about the degree of the anteroposterior velar (APV), oropharyngeal lateral wall (OPLW), or tongue-base (BT) obstruction based on the velum, oropharynx, base of tongue, and epiglottis (VOTE) classification. The data of 25 patients (9 female (36%)) were retrospectively evaluated. The mean age at the date of implantation was 54.52 ± 9.61 years, and the mean BMI was 29.99 ± 3.97 kg/m2. Spearman's rho correlation coefficients were calculated. Significant correlations were found between the degree of APV obstruction and postoperative HI (r = -0.5, p < 0.05), and between the degree of OPLW obstruction and postoperative snoring index (r = 0.42, p < 0.05). BT obstruction was strongly correlated with postoperative metrics such as AHI (r = -0.57, p < 0.01), AI (r = -0.5, p < 0.05), ODI (r = -0.57, p < 0.01), ∆ AHI (r = 0.58, p < 0.01), ∆ AI (r = 0.54, p < 0.01) and ∆ ODI (r = 0.54, p < 0.01). No significant correlation was found between DISE findings and postoperative Epworth Sleepiness Scale values. These findings suggest that preoperative DISE findings, especially the degree of BT obstruction, are important for predicting an HGNS therapy outcome.

Tinnitus Prevalence in the Adult Population-Results from the Gutenberg Health Study.

Background and Objectives: Tinnitus is a common symptom in medical practice, although data on its prevalence vary. As the underlying pathophysiological mechanism is still not fully understood, hearing loss is thought to be an important risk factor for the occurrence of tinnitus. The aim of this study was to assess tinnitus prevalence in a large German cohort and to determine its dependence on hearing impairment. Materials and Methods: The Gutenberg Health Study (GHS) is a population-based cohort study and representative for the population of Mainz and its district. Participants were asked whether they suffer from tinnitus and how much they are burdened by it. Extensive audiological examinations using bone- and air-conduction were also performed. Results: 4942 participants (mean age: 61.0, 2550 men and 2392 women) were included in the study. The overall prevalence of tinnitus was 26.1%. Men were affected significantly more often than women. The prevalence of tinnitus increased with age, peaking at ages 75 to 79 years. Considering only annoying tinnitus, the prevalence was 9.8%. Logistic regression showed that participants with severe to complete hearing loss (>65 dB) were more likely to have tinnitus. Conclusions: Tinnitus is a common symptom, and given demographic changes, its prevalence is expected to increase.

Central Apneic Event Prevalence in REM and NREM Sleep in OSA Patients: A Retrospective, Exploratory Study.

Patients with sleep-disordered breathing show a combination of different respiratory events (central, obstructive, mixed), with one type being predominant. We observed a reduced prevalence of central apneic events (CAEs) during REM sleep compared to NREM sleep in patients with predominant obstructive sleep apnea (OSA). The aim of this retrospective, exploratory study was to describe this finding and to suggest pathophysiological explanations. The polysomnography (PSG) data of 141 OSA patients were assessed for the prevalence of CAEs during REM and NREM sleep. On the basis of the apnea-hypopnea index (AHI), patients were divided into three OSA severity groups (mild: AHI < 15/h; moderate: AHI = 15-30/h; severe: AHI > 30/h). We compared the frequency of CAEs adjusted for the relative length of REM and NREM sleep time, and a significantly increased frequency of CAEs in NREM was found only in severely affected OSA patients. Given that the emergence of CAEs is strongly associated with the chemosensitivity of the brainstem nuclei regulating breathing mechanics in humans, a sleep-stage-dependent chemosensitivity is proposed. REM-sleep-associated neuronal circuits in humans may act protectively against the emergence of CAEs, possibly by reducing chemosensitivity. On the contrary, a significant increase in the chemosensitivity of the brainstem nuclei during NREM sleep is suggested.

Insomnia in Patients Undergoing Hypoglossal Nerve Stimulation Therapy for Obstructive Sleep Apnea.

Hypoglossal nerve stimulation (HGNS) is a treatment for obstructive sleep apnea (OSA) patients with intolerance of positive airway pressure therapy. Comorbid insomnia is quite prevalent in OSA patients. We investigated the impact of insomnia and excessive daytime sleepiness (EDS) on polysomnography metrics after HGNS treatment. Data of 27 consecutive patients (9 female; mean age 55.52 ± 8.6 years) were retrospectively evaluated. Insomnia was assessed using the ISI (insomnia severity index) and EDS using the Epworth sleepiness scale (ESS). The median ISI was reduced significantly 3 months after HGNS activation (preoperative: 19; postoperative: 14; p < 0.01). Significant correlations emerged between preoperative ISI and postoperative AHI (apnea/hypopnea index; Spearman's rho = 0.4, p < 0.05), ∆ AHI (r = −0.51, p < 0.01) and ∆ ODI (oxygen desaturation index; r = −0.48, p < 0.05). ISI correlated strongly with EES both preoperatively (r = 0.46; p < 0.02) and postoperatively (r = 0.79; p < 0.001). Therefore, HGNS therapy is associated with a significant reduction of insomnia-related symptoms, in addition to the improvement in respiratory metrics in OSA. Nonetheless, the preoperative severity of patient-reported insomnia symptoms was inversely correlated with the respiratory PSG-outcomes after HGNS. Insomnia should be considered in studies of EDS in OSA patients, especially those treated with HGNS.

Sensorimotor Cortical Activity during Respiratory Arousals in Obstructive Sleep Apnea.

Intensity of respiratory cortical arousals (RCA) is a pathophysiologic trait in obstructive sleep apnea (OSA) patients. We investigated the brain oscillatory features related to respiratory arousals in moderate and severe OSA. Raw electroencephalography (EEG) data recorded during polysomnography (PSG) of 102 OSA patients (32 females, mean age 51.6 ± 12 years) were retrospectively analyzed. Among all patients, 47 had moderate (respiratory distress index, RDI = 15−30/h) and 55 had severe (RDI > 30/h) OSA. Twenty RCA per sleep stage in each patient were randomly selected and a total of 10131 RCAs were analyzed. EEG signals obtained during, five seconds before and after the occurrence of each arousal were analyzed. The entropy (approximate (ApEn) and spectral (SpEn)) during each sleep stage (N1, N2 and REM) and area under the curve (AUC) of the EEG signal during the RCA was computed. Severe OSA compared to moderate OSA patients showed a significant decrease (p < 0.0001) in the AUC of the EEG signal during the RCA. Similarly, a significant decrease in spectral entropy, both before and after the RCA was observed, was observed in severe OSA patients when compared to moderate OSA patients. Contrarily, the approximate entropy showed an inverse pattern. The highest increase in approximate entropy was found in sleep stage N1. In conclusion, the dynamic range of sensorimotor cortical activity during respiratory arousals is sleep-stage specific, dependent on the frequency of respiratory events and uncoupled from autonomic activation. These findings could be useful for differential diagnosis of severe OSA from moderate OSA.