Useful blunders: Can automated speech recognition errors improve downstream dementia classification?

OBJECTIVES: We aimed to investigate how errors from automatic speech recognition (ASR) systems affect dementia classification accuracy, specifically in the "Cookie Theft" picture description task. We aimed to assess whether imperfect ASR-generated transcripts could provide valuable information for distinguishing between language samples from cognitively healthy individuals and those with Alzheimer's disease (AD). METHODS: We conducted experiments using various ASR models, refining their transcripts with post-editing techniques. Both these imperfect ASR transcripts and manually transcribed ones were used as inputs for the downstream dementia classification. We conducted comprehensive error analysis to compare model performance and assess ASR-generated transcript effectiveness in dementia classification. RESULTS: Imperfect ASR-generated transcripts surprisingly outperformed manual transcription for distinguishing between individuals with AD and those without in the "Cookie Theft" task. These ASR-based models surpassed the previous state-of-the-art approach, indicating that ASR errors may contain valuable cues related to dementia. The synergy between ASR and classification models improved overall accuracy in dementia classification. CONCLUSION: Imperfect ASR transcripts effectively capture linguistic anomalies linked to dementia, improving accuracy in classification tasks. This synergy between ASR and classification models underscores ASR's potential as a valuable tool in assessing cognitive impairment and related clinical applications.

A case study of Duchenne muscular dystrophy caused by Alu element insertion in DMD gene and analysis of its gray-hair symptoms.

Duchenne muscular dystrophy (DMD) is a severe X-linked recessive genetic disorder caused by mutations in the DMD gene, which leads to a deficiency of the dystrophin protein. The main mutation types of this gene include exon deletions and duplications, point mutations, and insertions. These mutations disrupt the normal expression of dystrophin, ultimately leading to the disease. In this study, we reported a case of DMD caused by an insertion mutation in exon 59 (E59) of the DMD gene. The affected child exhibited significant abnormalities in related biochemical markers, early symptoms of DMD, and multiple gray hair. His mother and sister were carriers with slightly abnormal biochemical markers. The mother had mild clinical symptoms, while the sister had no clinical symptoms. Other family members were genetically and physically normal. Sequencing and sequence alignment revealed that the inserted fragment was an Alu element from the AluYa5 subfamily. This insertion produced two stop codons and a polyadenylate (polyA) tail. To understand the impact of this insertion on the DMD gene and its association with clinical symptoms, exonic splicing enhancer (ESE) prediction indicated that the insertion did not affect the splicing of E59. Therefore, we speculated that the insertion sequence would be present in the mRNA sequence of the DMD gene. The two stop codons and polyA tail likely terminate translation, preventing the production of functional dystrophin protein, which may be the mechanism leading to DMD. In addition to typical DMD symptoms, the child also exhibited premature graying of hair. This study reports, for the first time, a case of DMD caused by the insertion of an Alu element into the coding region of the DMD gene. This finding provides clues for studying gene mutations induced by Alu sequence insertion and expands the understanding of DMD gene mutations.

A curious case of retrogenesis in language: Automated analysis of language patterns observed in dementia patients and young children.

Introduction: While linguistic retrogenesis has been extensively investigated in the neuroscientific and behavioral literature, there has been little work on retrogenesis using computerized approaches to language analysis. Methods: We bridge this gap by introducing a method based on comparing output of a pre-trained neural language model (NLM) with an artificially degraded version of itself to examine the transcripts of speech produced by seniors with and without dementia and healthy children during spontaneous language tasks. We compare a range of linguistic characteristics including language model perplexity, syntactic complexity, lexical frequency and part-of-speech use across these groups. Results: Our results indicate that healthy seniors and children older than 8 years share similar linguistic characteristics, as do dementia patients and children who are younger than 8 years. Discussion: Our study aligns with the growing evidence that language deterioration in dementia mirrors language acquisition in development using computational linguistic methods based on NLMs. This insight underscores the importance of further research to refine its application in guiding developmentally appropriate patient care, particularly in early stages.

Caffeine Ameliorates Age-Related Hearing Loss by Downregulating the Inflammatory Pathway in Mice.

OBJECTIVE: Age-related hearing loss (ARHL), also known as presbycusis, is a debilitating sensory impairment that affects the elderly population. There is currently no ideal treatment for ARHL. Long-term caffeine intake was reported to have anti-aging effects in many diseases. This study is to identify whether caffeine could ameliorate ARHL in mice and analyze its mechanism. METHODS: Caffeine was administered in drinking water to C57BL/6J mice from the age of 3 months to 12 months. The body weight, food intake and water intake of the mice were monitored during the experiment. The metabolic indicators of serum were detected by ELISA. The function of the hearing system was evaluated by ABR and hematoxylin and eosin staining of the cochlea. Genes' expression were detected by Q-PCR, immunofluorescencee and Western blot. RESULTS: The results showed that the ARHL mice exhibited impaired hearing and cochlear tissue compared with the young mice. However, the caffeine-treated ARHL mice showed improved hearing and cochlear tissue morphology. The expression of inflammation-related genes, such as TLR4, Myd88, NF-κB, and IL-1ß, was significantly increased in the cochleae of ARHL mice compared with young mice but was down-regulated in the caffeine-treated cochleae. CONCLUSIONS: Inflammation is involved in ARHL of mice, and long-term caffeine supplementation could ameliorate ARHL through the down-regulation of the TLR4/NF-κB inflammation pathway. Our findings provide a new idea for preventing ARHL and suggest new drug targets for ARHL treatment.

TRESTLE: Toolkit for Reproducible Execution of Speech, Text and Language Experiments.

The evidence is growing that machine and deep learning methods can learn the subtle differences between the language produced by people with various forms of cognitive impairment such as dementia and cognitively healthy individuals. Valuable public data repositories such as TalkBank have made it possible for researchers in the computational community to join forces and learn from each other to make significant advances in this area. However, due to variability in approaches and data selection strategies used by various researchers, results obtained by different groups have been difficult to compare directly. In this paper, we present TRESTLE (Toolkit for Reproducible Execution of Speech Text and Language Experiments), an open source platform that focuses on two datasets from the TalkBank repository with dementia detection as an illustrative domain. Successfully deployed in the hackallenge (Hackathon/Challenge) of the International Workshop on Health Intelligence at AAAI 2022, TRESTLE provides a precise digital blueprint of the data pre-processing and selection strategies that can be reused via TRESTLE by other researchers seeking comparable results with their peers and current state-of-the-art (SOTA) approaches.

Crosstalk of mRNA, miRNA, lncRNA, and circRNA and Their Regulatory Pattern in Pulmonary Fibrosis.

Noncoding RNAs (ncRNAs), such as microRNA (miRNA), long ncRNA (lncRNA), and circular RNA (circRNA), are regulators of important biological functions. Therefore, understanding their crosstalk and regulatory patterns can provide treatment for diseases. In this study, differentially expressed RNA transcripts were obtained by RNA sequencing in bleomycin-induced pulmonary fibrosis in mice. Four miRNAs, 10 lncRNAs, and two circRNAs were tested to validate the sequencing. There were differentially expressed 585 mRNAs, 236 miRNAs, 272 lncRNAs, and 74 circRNAs in pulmonary fibrosis. Their location on chromosome, length varieties, interaction, and host genes were analyzed. lnc949, circ949, and circ057 were chosen to explore the detailed crosstalk and regulatory pattern, which were measured by using RNA-FISH, dual-luciferase reporter assay, real-time cell analysis and rescue experiment, co-localization analysis, RNA immunoprecipitation, and RNA pull down. The data showed that the three ncRNAs were predominant in the cytoplasm, and their regulatory patterns were focused on post-transcription. The fibrotic function of lnc949 depended on its host gene FKBP5. circ949 and circ057 formed a regulatory network with lnc865 and lnc556 to simultaneously regulate miR-29b-2-5p targeting STAT3 phosphorylation. Collectively, different RNAs can crosstalk with each other to regulate pulmonary fibrosis through different regulatory patterns. We hope these data can provide a full concept of RNA transcripts, leading to a new treatment for pulmonary fibrosis.