Machine learning for risk stratification in the emergency department (MARS-ED) study protocol for a randomized controlled pilot trial on the implementation of a prediction model based on machine learning technology predicting 31-day mortality in the emergency department.

BACKGROUND: Many prediction models have been developed to help identify emergency department (ED) patients at high risk of poor outcome. However, these models often underperform in clinical practice and their actual clinical impact has hardly ever been evaluated. We aim to perform a clinical trial to investigate the clinical impact of a prediction model based on machine learning (ML) technology. METHODS: The study is a prospective, randomized, open-label, non-inferiority pilot clinical trial. We will investigate the clinical impact of a prediction model based on ML technology, the RISKINDEX, which has been developed to predict the risk of 31-day mortality based on the results of laboratory tests and demographic characteristics. In previous studies, the RISKINDEX was shown to outperform internal medicine specialists and to have high discriminatory performance. Adults patients (18 years or older) will be recruited in the ED. All participants will be randomly assigned to the control group or the intervention group in a 1:1 ratio. Participants in the control group will receive care as usual in which the study team asks the attending physicians questions about their clinical intuition. Participants in the intervention group will also receive care as usual, but in addition to asking the clinical impression questions, the study team presents the RISKINDEX to the attending physician in order to assess the extent to which clinical treatment is influenced by the results. DISCUSSION: This pilot clinical trial investigates the clinical impact and implementation of an ML based prediction model in the ED. By assessing the clinical impact and prognostic accuracy of the RISKINDEX, this study aims to contribute valuable insights to optimize patient care and inform future research in the field of ML based clinical prediction models. TRIAL REGISTRATION: ClinicalTrials.gov NCT05497830. Machine Learning for Risk Stratification in the Emergency Department (MARS-ED). Registered on August 11, 2022. URL: https://clinicaltrials.gov/study/NCT05497830 .

Imaging-based frequency mapping for cochlear implants - Evaluated using a daily randomized controlled trial.

Background: Due to variation in electrode design, insertion depth and cochlear morphology, patients with a cochlear implant (CI) often have to adapt to a substantial mismatch between the characteristic response frequencies of cochlear neurons and the stimulus frequencies assigned to electrode contacts. We introduce an imaging-based fitting intervention, which aimed to reduce frequency-to-place mismatch by aligning frequency mapping with the tonotopic position of electrodes. Results were evaluated in a novel trial set-up where subjects crossed over between intervention and control using a daily within-patient randomized approach, immediately from the start of CI rehabilitation. Methods: Fourteen adult participants were included in this single-blinded, daily randomized clinical trial. Based on a fusion of pre-operative imaging and a post-operative cone beam CT scan (CBCT), mapping of electrical input was aligned to natural place-pitch arrangement in the individual cochlea. That is, adjustments to the CI's frequency allocation table were made so electrical stimulation of frequencies matched as closely as possible with corresponding acoustic locations in the cochlea. For a period of three months, starting at first fit, a scheme was implemented whereby the blinded subject crossed over between the experimental and standard fitting program using a daily randomized wearing schedule, and thus effectively acted as their own control. Speech outcomes (such as speech intelligibility in quiet and noise, sound quality and listening effort) were measured with both settings throughout the study period. Results: On a group level, standard fitting obtained subject preference and showed superior results in all outcome measures. In contrast, two out of fourteen subjects preferred the imaging-based fitting and correspondingly had better speech understanding with this setting compared to standard fitting. Conclusion: On average, cochlear implant fitting based on individual tonotopy did not elicit higher speech intelligibility but variability in individual results strengthen the potential for individualized frequency fitting. The novel trial design proved to be a suitable method for evaluation of experimental interventions in a prospective trial setup with cochlear implants.

Extending the audiogram with loudness growth: The complementarity of electric and acoustic hearing in bimodal patients.

INTRODUCTION: Clinically, recording hearing detection thresholds and representing them in an audiogram is the most common way of evaluating hearing loss and starting the fitting of hearing devices. As an extension, we present the loudness audiogram, which does not only show auditory thresholds but also visualizes the full course of loudness growth across frequencies. The benefit of this approach was evaluated in subjects who rely on both electric (cochlear implant) and acoustic (hearing aid) hearing. METHODS: In a group of 15 bimodal users, loudness growth was measured with the cochlear implant and hearing aid separately using a loudness scaling procedure. Loudness growth curves were constructed, using a novel loudness function, for each modality and then integrated in a graph plotting frequency, stimulus intensity level, and loudness perception. Bimodal benefit, defined as the difference between wearing a cochlear implant and hearing aid together versus wearing only a cochlear implant, was assessed for multiple speech outcomes. RESULTS: Loudness growth was related to bimodal benefit for speech recognition in noise and to some aspects of speech quality. No correlations between loudness and speech in quiet were found. Patients who had predominantly unequal loudness input from the hearing aid, gained more bimodal benefit for speech recognition in noise compared to those patients whose hearing aid provided mainly equivalent input. CONCLUSION: Results show that loudness growth is related to bimodal benefit for speech recognition in noise and to some aspects of speech quality. Subjects who had different input from the hearing aid compared to CI, generally gained more bimodal benefit compared to those patients whose hearing aid provided mainly equivalent input. This suggests that bimodal fitting to create equal loudness at all frequencies may not always be beneficial for speech recognition.

Toward neural health measurements for cochlear implantation: The relationship among electrode positioning, the electrically evoked action potential, impedances and behavioral stimulation levels.

Introduction: Estimating differences in neural health across different sites within the individual cochlea potentially enables clinical applications for subjects with a cochlear implant. The electrically evoked compound action potential (ECAP) is a measure of neural excitability that possibly provides an indication of a neural condition. There are many factors, however, that affect this measure and increase the uncertainty of its interpretation. To better characterize the ECAP response, its relationship with electrode positioning, impedances, and behavioral stimulation levels was explored. Methods: A total of 14 adult subjects implanted with an Advanced Bionics cochlear electrode array were prospectively followed up from surgery to 6 months postoperative. Insertion depth, distance to the modiolus, and distance to the medial wall were assessed for each electrode by postoperative CT analysis. ECAPs were measured intraoperatively and at three visits postoperatively on all 16 electrodes using the NRI feature of clinical programming software and characterized using multiple parameters. Impedances and behavioral stimulation levels were measured at every fitting session. Results: Patterns in ECAPs and impedances were consistent over time, but high variability existed among subjects and between different positions in the cochlea. Electrodes located closer to the apex of the cochlea and closer to the modiolus generally showed higher neural excitation and higher impedances. Maximum loudness comfort levels were correlated strongly with the level of current needed to elicit a response of 100 µV ECAP. Conclusion: Multiple factors contribute to the ECAP response in subjects with a cochlear implant. Further research might address whether the ECAP parameters used in this study will benefit clinical electrode fitting or the assessment of auditory neuron integrity.

CT in relation to RT-PCR in diagnosing COVID-19 in The Netherlands: A prospective study.

INTRODUCTION: Early differentiation between emergency department (ED) patients with and without corona virus disease (COVID-19) is very important. Chest CT scan may be helpful in early diagnosing of COVID-19. We investigated the diagnostic accuracy of CT using RT-PCR for SARS-CoV-2 as reference standard and investigated reasons for discordant results between the two tests. METHODS: In this prospective single centre study in the Netherlands, all adult symptomatic ED patients had both a CT scan and a RT-PCR upon arrival at the ED. CT results were compared with PCR test(s). Diagnostic accuracy was calculated. Discordant results were investigated using discharge diagnoses. RESULTS: Between March 13th and March 24th 2020, 193 symptomatic ED patients were included. In total, 43.0% of patients had a positive PCR and 56.5% a positive CT, resulting in a sensitivity of 89.2%, specificity 68.2%, likelihood ratio (LR)+ 2.81 and LR- 0.16. Sensitivity was higher in patients with high risk pneumonia (CURB-65 score ≥3; n = 17, 100%) and with sepsis (SOFA score ≥2; n = 137, 95.5%). Of the 35 patients (31.8%) with a suspicious CT and a negative RT-PCR, 9 had another respiratory viral pathogen, and in 7 patients, COVID-19 was considered likely. One of nine patients with a non-suspicious CT and a positive PCR had developed symptoms within 48 hours before scanning. DISCUSSION: The accuracy of chest CT in symptomatic ED patients is high, but used as a single diagnostic test, CT can not safely diagnose or exclude COVID-19. However, CT can be used as a quick tool to categorize patients into "probably positive" and "probably negative" cohorts.

Energy Expenditure during Extreme Endurance Exercise: The Giro d'Italia.

PURPOSE: Little data are available on doubly labeled water (DLW) assessed total daily energy expenditure (TDEE) during extreme endurance exercise. Doubly labeled water is considered the gold standard to measure TDEE, but different calculations are being used, which may have a large impact on the results. The aim of the current study was to measure TDEE during the Giro d'Italia and apply two different calculation methods. METHODS: Seven male cyclists (age, 28 ± 5 yr; body mass index, 22.1 ± 2.1 kg·m) completed the 24-d professional cycling race "Giro d'Italia" in which a total distance of 3445 km was covered, including 10 mountain stages. Total daily energy expenditure was measured over the entire duration of the race, with the ingestion of DLW at three different time points. To calculate TDEE and body composition, the isotope dilution space was calculated using two different techniques, the "plateau" and "intercept" technique. RESULTS: The %fat mass at baseline was 7.8% and 16.8% with the plateau and intercept technique respectively and did not significantly change over the course of the race. Total daily energy expenditure was on average 32.3 ± 3.4 MJ·d using the plateau technique versus 28.9 ± 3.2 using the intercept technique, resulting in an average physical activity level (PAL) of 4.37 ± 0.43 versus 3.91 ± 0.39, respectively. The dilution space ratio was on average 1.030 with the plateau and 1.060 with the intercept technique. CONCLUSIONS: Given that the observed dilution space ratio with the plateau technique is similar as the expected ratio from literature and the % fat mass of 7.8% is more realistic for the athletes being studied, we propose the application of the plateau rather than the intercept method, when using DLW during extreme endurance exercise.