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 the limitations of speech-based systems, 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 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 modeled speech with pretrained deep learning models that were pretrained on a large, open, and free database, and we selected the best one on the validation set. Based on the best speech modeling approach, clinical threshold detection, individual score prediction, model uncertainty estimation, and performance fairness across demographics (age, sex, and education) were evaluated. We used 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 Whisper M with a max pooling and oversampling method. Our methods achieved good detection performance for all symptoms, depression (Patient Health Questionnaire-9: area under the curve [AUC]=0.76; F1-score=0.49 and Beck Depression Inventory: AUC=0.78; F1-score=0.65), anxiety (Generalized Anxiety Disorder 7-item scale: AUC=0.77; F1-score=0.50), insomnia (Athens Insomnia Scale: AUC=0.73; F1-score=0.62), and fatigue (Multidimensional Fatigue Inventory total score: AUC=0.68; F1-score=0.88). The system performed well when it needed to abstain from making predictions, as demonstrated by low abstention rates in depression detection with the Beck Depression Inventory and fatigue, with risk-coverage AUCs below 0.4. Individual symptom scores were accurately predicted (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, SD 0.13), to a lesser extent for education level (average DR 0.47, SD 0.30), and worse for age groups (average DR 0.33, SD 0.30). 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.

Erythroid-Progenitor-Targeted Gene Therapy Using Bifunctional TFR1 Ligand-Peptides in Human Erythropoietic Protoporphyria.

Erythropoietic protoporphyria (EPP) is a hereditary disease characterized by a deficiency in ferrochelatase (FECH) activity. FECH activity is responsible for the accumulation of protoporphyrin IX (PPIX). Without etiopathogenic treatment, EPP manifests as severe photosensitivity. 95% of affected individuals present a hypomorphic FECH allele trans to a loss-of-function (LOF) FECH mutation, resulting in a reduction in FECH activity in erythroblasts below a critical threshold. The hypomorphic allele promotes the use of a cryptic acceptor splice site, generating an aberrant FECH mRNA, which is responsible for the reduced level of wild-type FECH mRNA and, ultimately, FECH activity. We have previously identified an antisense oligonucleotide (AON), AON-V1 (V1), that redirects splicing to the physiological acceptor site and reduces the accumulation of PPIX. Here, we developed a specific strategy that uses transferrin receptor 1 (TRF1) as a Trojan horse to deliver V1 to erythroid progenitors. We designed a bifunctional peptide (P1-9R) including a TFR1-targeting peptide coupled to a nine-arginine cell-penetrating peptide (CPP) that facilitates the release of the AON from TFR1 in endosomal vesicles. We demonstrated that the P1-9R/V1 nanocomplex promotes the efficient and prolonged redirection of splicing towards the physiological splice site and subsequent normalization of WT FECH mRNA and protein levels. Finally, the P1-9R/V1 nanocomplex increases WT FECH mRNA production and significantly decreases PPIX accumulation in primary cultures of differentiating erythroid progenitors from an overt EPP-affected individual. P1-9R is a method designed to target erythroid progenitors and represents a potentially powerful tool for the in vivo delivery of therapeutic DNA in many erythroid disorders.

Antisense oligonucleotide-based therapy in human erythropoietic protoporphyria.

In 90% of people with erythropoietic protoporphyria (EPP), the disease results from the inheritance of a common hypomorphic FECH allele, encoding ferrochelatase, in trans to a private deleterious FECH mutation. The activity of the resulting FECH enzyme falls below the critical threshold of 35%, leading to the accumulation of free protoporphyrin IX (PPIX) in bone marrow erythroblasts and in red cells. The mechanism of low expression involves a biallelic polymorphism (c.315-48T>C) localized in intron 3. The 315-48C allele increases usage of the 3' cryptic splice site between exons 3 and 4, resulting in the transcription of an unstable mRNA with a premature stop codon, reducing the abundance of wild-type FECH mRNA, and finally reducing FECH activity. Through a candidate-sequence approach and an antisense-oligonucleotide-tiling method, we identified a sequence that, when targeted by an antisense oligonucleotide (ASO-V1), prevented usage of the cryptic splice site. In lymphoblastoid cell lines derived from symptomatic EPP subjects, transfection of ASO-V1 reduced the usage of the cryptic splice site and efficiently redirected the splicing of intron 3 toward the physiological acceptor site, thereby increasing the amount of functional FECH mRNA. Moreover, the administration of ASO-V1 into developing human erythroblasts from an overtly EPP subject markedly increased the production of WT FECH mRNA and reduced the accumulation of PPIX to a level similar to that measured in asymptomatic EPP subjects. Thus, EPP is a paradigmatic Mendelian disease in which the in vivo correction of a common single splicing defect would improve the condition of most affected individuals.

Epistasis in iron metabolism: complex interactions between Cp, Mon1a, and Slc40a1 loci and tissue iron in mice.

Disorders of iron metabolism are among the most common acquired and constitutive diseases. Hemochromatosis has a solid genetic basis and in Northern European populations it is usually associated with homozygosity for the C282Y mutation in the HFE protein. However, the penetrance of this mutation is incomplete and the clinical presentation is highly variable. The rare and common variants identified so far as genetic modifiers of HFE-related hemochromatosis are unable to account for the phenotypic heterogeneity of this disorder. There are wide variations in the basal iron status of common inbred mouse strains, and this diversity may reflect the genetic background of the phenotypic diversity under pathological conditions. We therefore examined the genetic basis of iron homeostasis using quantitative trait loci mapping applied to the HcB-15 recombinant congenic strains for tissue and serum iron indices. Two highly significant QTL containing either the N374S Mon1a mutation or the Ferroportin locus were found to be major determinants in spleen and liver iron loading. Interestingly, when considering possible epistatic interactions, the effects of Mon1a on macrophage iron export are conditioned by the genotype at the Slc40a1 locus. Only mice that are C57BL/10ScSnA homozygous at both loci display a lower spleen iron burden. Furthermore, the liver-iron lowering effect of the N374S Mon1a mutation is observed only in mice that display a nonsense mutation in the Ceruloplasmin (Cp) gene. This study highlights the existence of genetic interactions between Cp, Mon1a, and the Slc40a1 locus in iron metabolism, suggesting that epistasis may be a crucial determinant of the variable biological and clinical presentations in iron disorders.

Protoporphyrin retention in hepatocytes and Kupffer cells prevents sclerosing cholangitis in erythropoietic protoporphyria mouse model.

BACKGROUND & AIMS: Chronic, progressive hepatobiliary disease is the most severe complication of erythropoietic protoporphyria (EPP) and can require liver transplantation, although the mechanisms that lead to liver failure are unknown. We characterized protoporphyrin-IX (PPIX)-linked hepatobiliary disease in BALB/c and C57BL/6 (Fechm1Pas) mice with mutations in ferrochelatase as models for EPP. METHODS: Fechm1Pas and wild-type (control) mice were studied at 12-14 weeks of age. PPIX was quantified; its distribution in the liver, serum levels of lipoprotein-X, liver histology, contents of bile salt and cholesterol phospholipids, and expression of genes were compared in mice of the BALB/c and C57BL/6 backgrounds. The in vitro binding affinity of PPIX for bile components was determined. RESULTS: Compared with mice of the C57BL/6 background, BALB/c Fechm1Pas mice had a more severe pattern of cholestasis, fibrosis with portoportal bridging, bile acid regurgitation, sclerosing cholangitis, and hepatolithiasis. In C57BL/6 Fechm1Pas mice, PPIX was sequestrated mainly in the cytosol of hepatocytes and Kupffer cells, whereas, in BALB/c Fechm1Pas mice, PPIX was localized within enlarged bile canaliculi. Livers of C57BL/6 Fechm1Pas mice were protected through a combination of lower efflux of PPIX and reduced synthesis and export of bile acid. CONCLUSIONS: PPIX binds to bile components and disrupts the physiologic equilibrium of phospholipids, bile acids, and cholesterol in bile. This process might be involved in pathogenesis of sclerosing cholangitis from EPP; a better understanding might improve diagnosis and development of reagents to treat or prevent liver failure in patients with EPP.