Automated speech analysis for risk detection of depression, anxiety, insomnia, and fatigue: Algorithm Development and Validation Study.

BACKGROUND: While speech analysis holds promise for mental health assessment, research often focuses on single symptoms, despite symptom co-occurrences and interactions. In addition, predictive models in mental health do not properly assess speech-based systems' limitations, such as uncertainty, or fairness for a safe clinical deployment. OBJECTIVE: We investigated the predictive potential of mobile-collected speech data for detecting and estimating depression, anxiety, fatigue, and insomnia, focusing on other factors than mere accuracy, in the general population. METHODS: We included n=865 healthy adults and recorded their answers regarding their perceived mental and sleep states. We asked how they felt and if they had slept well lately. Clinically validated questionnaires measuring depression, anxiety, insomnia, and fatigue severity were also used. We developed a novel speech and machine learning pipeline involving voice activity detection, feature extraction, and model training. We automatically analyzed participants' speech with a fully ML automatic pipeline to capture speech variability. Then, we modelled speech with pretrained deep learning models that were pre-trained on a large open free database and we selected the best one on the validation set. Based on the best speech modelling approach, we evaluated clinical threshold detection, individual score prediction, model uncertainty estimation, and performance fairness across demographics (age, sex, education). We employed a train-validation-test split for all evaluations: to develop our models, select the best ones and assess the generalizability of held-out data. RESULTS: The best model was WhisperM with a max pooling, and oversampling method. Our methods achieved good detection performance for all symptoms, depression (PHQ-9 AUC= 0.76F1=0.49, BDI AUC=0.78, F1=0,65), anxiety (GAD-7 F1=0.50, AUC=0.77) insomnia (AIS AUC=0.73, F1=0.62), and fatigue (MFI Total Score F1=0.88, AUC=0.68). These strengths were maintained for depression detection with BDI and Fatigue for abstention rates for uncertain cases (Risk-Coverage AUCs < 0.4). Individual symptom scores were predicted with good accuracy (Correlations were all significant, with Pearson strengths between 0.31 and 0.49). Fairness analysis revealed that models were consistent for sex (average Disparity Ratio (DR) = 0.86), to a lesser extent for education level (average Disparity Ratio (DR) = 0.47) and worse for age groups (average Disparity Ratio (DR) = 0.33). CONCLUSIONS: This study demonstrates the potential of speech-based systems for multifaceted mental health assessment in the general population, not only for detecting clinical thresholds but also for estimating their severity. Addressing fairness and incorporating uncertainty estimation with selective classification are key contributions that can enhance the clinical utility and responsible implementation of such systems. This approach offers promise for more accurate and nuanced mental health assessments, benefiting both patients and clinicians.

FHF2 phosphorylation and regulation of native myocardial Na V 1.5 channels.

Phosphorylation of the cardiac Na V 1.5 channel pore-forming subunit is extensive and critical in modulating channel expression and function, yet the regulation of Na V 1.5 by phosphorylation of its accessory proteins remains elusive. Using a phosphoproteomic analysis of Na V channel complexes purified from mouse left ventricles, we identified nine phosphorylation sites on Fibroblast growth factor Homologous Factor 2 (FHF2). To determine the roles of phosphosites in regulating Na V 1.5, we developed two models from neonatal and adult mouse ventricular cardiomyocytes in which FHF2 expression is knockdown and rescued by WT, phosphosilent or phosphomimetic FHF2-VY. While the increased rates of closed-state and open-state inactivation of Na V channels induced by the FHF2 knockdown are completely restored by the FHF2-VY isoform in adult cardiomyocytes, sole a partial rescue is obtained in neonatal cardiomyocytes. The FHF2 knockdown also shifts the voltage-dependence of activation towards hyperpolarized potentials in neonatal cardiomyocytes, which is not rescued by FHF2-VY. Parallel investigations showed that the FHF2-VY isoform is predominant in adult cardiomyocytes, while expression of FHF2-VY and FHF2-A is comparable in neonatal cardiomyocytes. Similar to WT FHF2-VY, however, each FHF2-VY phosphomutant restores the Na V channel inactivation properties in both models, preventing identification of FHF2 phosphosite roles. FHF2 knockdown also increases the late Na + current in adult cardiomyocytes, which is restored similarly by WT and phosphosilent FHF2-VY. Together, our results demonstrate that ventricular FHF2 is highly phosphorylated, implicate differential roles for FHF2 in regulating neonatal and adult mouse ventricular Na V 1.5, and suggest that the regulation of Na V 1.5 by FHF2 phosphorylation is highly complex. eTOC Summary: Lesage et al . identify the phosphorylation sites of FHF2 from mouse left ventricular Na V 1.5 channel complexes. While no roles for FHF2 phosphosites could be recognized yet, the findings demonstrate differential FHF2-dependent regulation of neonatal and adult mouse ventricular Na V 1.5 channels.

Determinants of iFGF13-mediated regulation of myocardial voltage-gated sodium (NaV) channels in mouse.

Posttranslational regulation of cardiac NaV1.5 channels is critical in modulating channel expression and function, yet their regulation by phosphorylation of accessory proteins has gone largely unexplored. Using phosphoproteomic analysis of NaV channel complexes from adult mouse left ventricles, we identified nine phosphorylation sites on intracellular fibroblast growth factor 13 (iFGF13). To explore the potential roles of these phosphosites in regulating cardiac NaV currents, we abolished expression of iFGF13 in neonatal and adult mouse ventricular myocytes and rescued it with wild-type (WT), phosphosilent, or phosphomimetic iFGF13-VY. While the increased rate of closed-state inactivation of NaV channels induced by Fgf13 knockout in adult cardiomyocytes was completely restored by adenoviral-mediated expression of WT iFGF13-VY, only partial rescue was observed in neonatal cardiomyocytes after knockdown. The knockdown of iFGF13 in neonatal ventricular myocytes also shifted the voltage dependence of channel activation toward hyperpolarized potentials, a shift that was not reversed by WT iFGF13-VY expression. Additionally, we found that iFGF13-VY is the predominant isoform in adult ventricular myocytes, whereas both iFGF13-VY and iFGF13-S are expressed comparably in neonatal ventricular myocytes. Similar to WT iFGF13-VY, each of the iFGF13-VY phosphomutants studied restored NaV channel inactivation properties in both models. Lastly, Fgf13 knockout also increased the late Na+ current in adult cardiomyocytes, and this effect was restored with expression of WT and phosphosilent iFGF13-VY. Together, our results demonstrate that iFGF13 is highly phosphorylated and displays differential isoform expression in neonatal and adult ventricular myocytes. While we found no roles for iFGF13 phosphorylation, our results demonstrate differential effects of iFGF13 on neonatal and adult mouse ventricular NaV channels.

Computer modeling of whole-cell voltage-clamp analyses to delineate guidelines for good practice of manual and automated patch-clamp.

The patch-clamp technique and more recently the high throughput patch-clamp technique have contributed to major advances in the characterization of ion channels. However, the whole-cell voltage-clamp technique presents certain limits that need to be considered for robust data generation. One major caveat is that increasing current amplitude profoundly impacts the accuracy of the biophysical analyses of macroscopic ion currents under study. Using mathematical kinetic models of a cardiac voltage-gated sodium channel and a cardiac voltage-gated potassium channel, we demonstrated how large current amplitude and series resistance artefacts induce an undetected alteration in the actual membrane potential and affect the characterization of voltage-dependent activation and inactivation processes. We also computed how dose-response curves are hindered by high current amplitudes. This is of high interest since stable cell lines frequently demonstrating high current amplitudes are used for safety pharmacology using the high throughput patch-clamp technique. It is therefore critical to set experimental limits for current amplitude recordings to prevent inaccuracy in the characterization of channel properties or drug activity, such limits being different from one channel type to another. Based on the predictions generated by the kinetic models, we draw simple guidelines for good practice of whole-cell voltage-clamp recordings.

Proteomic and functional mapping of cardiac NaV1.5 channel phosphorylation sites.

Phosphorylation of the voltage-gated Na+ (NaV) channel NaV1.5 regulates cardiac excitability, yet the phosphorylation sites regulating its function and the underlying mechanisms remain largely unknown. Using a systematic, quantitative phosphoproteomic approach, we analyzed NaV1.5 channel complexes purified from nonfailing and failing mouse left ventricles, and we identified 42 phosphorylation sites on NaV1.5. Most sites are clustered, and three of these clusters are highly phosphorylated. Analyses of phosphosilent and phosphomimetic NaV1.5 mutants revealed the roles of three phosphosites in regulating NaV1.5 channel expression and gating. The phosphorylated serines S664 and S667 regulate the voltage dependence of channel activation in a cumulative manner, whereas the nearby S671, the phosphorylation of which is increased in failing hearts, regulates cell surface NaV1.5 expression and peak Na+ current. No additional roles could be assigned to the other clusters of phosphosites. Taken together, our results demonstrate that ventricular NaV1.5 is highly phosphorylated and that the phosphorylation-dependent regulation of NaV1.5 channels is highly complex, site specific, and dynamic.

Smoothed Biasing Forces Yield Unbiased Free Energies with the Extended-System Adaptive Biasing Force Method.

We report a theoretical description and numerical tests of the extended-system adaptive biasing force method (eABF), together with an unbiased estimator of the free energy surface from eABF dynamics. Whereas the original ABF approach uses its running estimate of the free energy gradient as the adaptive biasing force, eABF is built on the idea that the exact free energy gradient is not necessary for efficient exploration, and that it is still possible to recover the exact free energy separately with an appropriate estimator. eABF does not directly bias the collective coordinates of interest, but rather fictitious variables that are harmonically coupled to them; therefore is does not require second derivative estimates, making it easily applicable to a wider range of problems than ABF. Furthermore, the extended variables present a smoother, coarse-grain-like sampling problem on a mollified free energy surface, leading to faster exploration and convergence. We also introduce CZAR, a simple, unbiased free energy estimator from eABF trajectories. eABF/CZAR converges to the physical free energy surface faster than standard ABF for a wide range of parameters.