Editorial: Towards an Understanding of Tinnitus Heterogeneity

This study delves into the multifaceted nature of cross-validation (CV) techniques in machine learning model evaluation and selection, underscoring the challenge of choosing the most appropriate method due to the plethora of available variants. It aims to clarify and standardize terminology such as sets, groups, folds, and samples pivotal in the CV domain, and introduces an exhaustive compilation of advanced CV methods like leave-one-out, leave-p-out, Monte Carlo, grouped, stratified, and time-split CV within a hold-out CV framework. Through graphical representations, the paper enhances the comprehension of these methodologies, facilitating more informed decision making for practitioners. It further explores the synergy between different CV strategies and advocates for a unified approach to reporting model performance by consolidating essential metrics. The paper culminates in a comprehensive overview of the CV techniques discussed, illustrated with practical examples, offering valuable insights for both novice and experienced researchers in the field.

Medical use cases for machine learning (ML) are growing exponentially. The first hospitals are already using ML systems as decision support systems in their daily routine. At the same time, most ML systems are still opaque and it is not clear how these systems arrive at their predictions. In this paper, we provide a brief overview of the taxonomy of explainability methods and review popular methods. In addition, we conduct a systematic literature search on PubMed to investigate which explainable artificial intelligence (XAI) methods are used in 450 specific medical supervised ML use cases, how the use of XAI methods has emerged recently, and how the precision of describing ML pipelines has evolved over the past 20 years. A large fraction of publications with ML use cases do not use XAI methods at all to explain ML predictions. However, when XAI methods are used, open-source and model-agnostic explanation methods are more commonly used, with SHapley Additive exPlanations (SHAP) and Gradient Class Activation Mapping (Grad-CAM) for tabular and image data leading the way. ML pipelines have been described in increasing detail and uniformity in recent years. However, the willingness to share data and code has stagnated at about one-quarter. XAI methods are mainly used when their application requires little effort. The homogenization of reports in ML use cases facilitates the comparability of work and should be advanced in the coming years. Experts who can mediate between the worlds of informatics and medicine will become more and more in demand when using ML systems due to the high complexity of the domain.

Background: The number of mobile health apps (MHAs), which are developed to promote healthy behaviors, prevent disease onset, manage and cure diseases, or assist with rehabilitation measures, has exploded. App store star ratings and descriptions usually provide insufficient or even false information about app quality, although they are popular among end users. A rigorous systematic approach to establish and evaluate the quality of MHAs is urgently needed. The Mobile App Rating Scale (MARS) is an assessment tool that facilitates the objective and systematic evaluation of the quality of MHAs. However, a German MARS is currently not available. Objective: The aim of this study was to translate and validate a German version of the MARS (MARS-G). Methods: The original 19-item MARS was forward and backward translated twice, and the MARS-G was created. App description items were extended, and 104 MHAs were rated twice by eight independent bilingual researchers, using the MARS-G and MARS. The internal consistency, validity, and reliability of both scales were assessed. Mokken scale analysis was used to investigate the scalability of the overall scores. Results: The retranslated scale showed excellent alignment with the original MARS. Additionally, the properties of the MARS-G were comparable to those of the original MARS. The internal consistency was good for all subscales (ie, omega ranged from 0.72 to 0.91). The correlation coefficients (r) between the dimensions of the MARS-G and MARS ranged from 0.93 to 0.98. The scalability of the MARS (H=0.50) and MARS-G (H=0.48) were good. Conclusions: The MARS-G is a reliable and valid tool for experts and stakeholders to assess the quality of health apps in German-speaking populations. The overall score is a reliable quality indicator. However, further studies are needed to assess the factorial structure of the MARS and MARS-G.

The psychological process how tinnitus loudness leads to tinnitus distress remains unclear. This cross-sectional study investigated the mediating role of the emotional state “stress level” and of the two components of the emotional state “arousal” and “valence” with N = 658 users of the “TrackYourTinnitus” smartphone application. Stress mediated the relationship between tinnitus loudness and tinnitus distress in a simple mediation model and even in a multiple mediation model when arousal and valence were held constant. Arousal mediated the loudness-distress relationship when holding valence constant, but not anymore when controlling for valence as well as for stress. Valence functioned as a mediator when controlling for arousal and even when holding arousal and stress constant. The direct effect of tinnitus loudness on tinnitus distress remained significant in all models. This study demonstrates that emotional states affect the process how tinnitus loudness leads to tinnitus distress. We thereby could show that the mediating influence of emotional valence is at least equally strong as the influence of stress. Implications of the findings for future research, assessment and clinical management of tinnitus are discussed.

Tinnitus, the phantom perception of sound without a corresponding external sound, is a frequent disorder which causes significant morbidity. So far there is no treatment available that reliably reduces the tinnitus perception. The research is hampered by the large heterogeneity of tinnitus and the fact that the tinnitus perception fluctuates over time. It is therefore necessary to develop tools for measuring fluctuations of tinnitus perception over time and for analysing data on single subject basis. However, this type of longitudinal measurement is difficult to perform using the traditional research methods such as paper-and-pencil questionnaires or clinical interviews. Ecological Momentary Assessment (EMA) represents a research concept that allows the assessment of subjective measurements under real-life conditions using portable electronic devices and thereby enables the researcher to collect longitudinal data under real-life conditions and high cost efficiency. Here we present a new method for recording the longitudinal development of tinnitus perception using a modern smartphone application available for iOS and Android devices with no costs for the users. The TrackYourTinnitus app is available and maintained since April 2014. A number of 857 volunteers with an average age of 44.1 years participated in the data collection between April 2014 and February 2016. The mean tinnitus distress at the initial measurement was rated on average 13.9 points on the Mini-Tinnitus Questionnaire (max. 24 points). Importantly, we could demonstrate that the regular use of the TrackYourTinnitus app has no significant negative influence on the perception of the tinnitus loudness nor on the tinnitus distress. The TrackYourTinnitus app can therefore be proposed as a safe instrument for the longitudinal assessment of tinnitus perception in the everyday life of the patient.