Brain Diseases Analysis Laboratory

Hypokinetic dysarthria (HD) is a motor speech disorder present in up to 90% of patients with Parkinson’s disease (PD). HD can manifest in the areas of respiration, phonation, articulation, and/or prosody, having a detrimental impact on patients' quality of life. Since the most common approach to PD treatment (i.e., based on L-DOPA) has a very individual effect on HD (likely dependent on the progression and subtype of PD), we have collaborated with neuroscientists, psychologists, and speech-language pathologists from the CEITEC - Central European Institute of Technology to explore new possibilities for improving speech in PD patients. For instance, within the framework of the project "Effects of non-invasive brain stimulation on hypokinetic dysarthria, micrographia, and brain plasticity in patients with Parkinson's disease" (https://lnkd.in/e-bATJfd), we have identified a beneficial effect of repetitive transcranial magnetic stimulation of the right superior temporal gyrus on articulation and other speech aspects: 1. https://lnkd.in/dhPSpEJc 2. https://lnkd.in/eMf_ukJF 3. https://lnkd.in/eG_yT4JU Subsequently, we became involved in a project (refer to https://lnkd.in/esGsYjhf) led by Luboš Brabenec, where we are currently supporting efforts to simplify the entire process, enabling patients to administer the stimulation themselves at home. In this case, we are utilizing the technology of transcranial direct-current stimulation in combination with the Lee Silverman Voice Treatment. Initial results will be published soon. Throughout this research, digital vocal biomarkers are extensively employed to facilitate neuroscientists in objectively monitoring the effects of new therapies and determining their optimal settings. If you are interested in applying digital vocal biomarkers in your own studies or clinical trials, we would be more than happy to assist you. Brno University of Technology #ParkinsonsDisease #dysarthria #treatment #rTMS #tDCS #DigitalBiomarkers #speech #voice #pathology

🔍 Thrilled to share insights from our recent study on the ethical dimensions of using voice as a health biomarker in AI! Conducted with the support of the Bridge2AI-Voice Consortium, this research highlights the diverse perspectives of stakeholders—including experts, clinicians, and patients—on the challenges and opportunities in developing trustworthy voice AI technologies. 📊 Key findings include: - Identification of ethical priorities for voice AI development - Initial definitions of ethically sourced data - Insights into the role of synthetic voice data By understanding these perspectives, we can work towards creating voice datasets that are free from bias, promoting inclusion, and addressing health disparities in marginalized communities. This study is a significant step forward in ensuring that voice AI technologies are both ethical and effective in health applications. For more details, check out the full study here: https://lnkd.in/eQ8d_dRS @jeanchristophebelislepipon @varditravitsky @mariapowell @yaelbensoussan @reneeenglish @mariefrancoisemalo #VoiceAI #HealthTech #EthicsInAI #Innovation #Bridge2AI

🌟 Exciting News from the Bridge2AI-Voice Consortium! 🌟 We are thrilled to announce the publication of our latest paper detailing our innovative protocols for data acquisition across multiple disease cohorts encompassing voice, respiratory, neurodegenerative diseases, and mood and anxiety disorders. Our comprehensive approach includes demographic surveys, confounder assessments, acoustic tasks, validated patient-reported outcome (PRO) questionnaires, and clinician-validated diagnostic questions, all designed to enhance the research community's capacity for broader adoption and valuable feedback. This collaborative effort brings together the expertise of numerous dedicated professionals and the Bridge2AI-Voice program, part of Bridge2AI, funded by the NIH Common Fund. As we harness the power of artificial intelligence (AI) to analyze voice, speech, and respiratory sound data, our project aims to create a diverse, ethically sourced voice database linked to health information, propelling Voice AI research into new frontiers. 🎤✨ Continue to follow us to learn more about our work, and to stay up to date on upcoming events from our consortium including data releases and our Voice AI Symposium. 🔗 doi: 10.21437/Interspeech.2024-1926 #Bridge2AI #VoiceAI #Audiomics #TeamScience #HealthcareInnovation #DEI #Bioethics #ClinicalResearch

Exciting advancements in voice data collection! As part of our ongoing efforts to enhance voice AI research, our latest study investigates how easily accessible iOS and Android tablets—both with and without low-cost headset microphones—can deliver recordings on par with traditional research-quality equipment. 🔍 Key findings reveal that recordings made with tablets and headset microphones positioned close to the mouth yielded highly accurate voice measurements, closely correlating with standard research methods. In contrast, using built-in tablet microphones at standard reading distances led to significant variability and less reliable data. 📊 These insights align with the Bridge2AI-Voice Consortium’s recommendations, highlighting the effectiveness of smartphones and headset microphones for large-scale voice data collection across clinical and nonclinical environments. This study opens new doors for practical, high-quality voice data collection, making it easier to fuel innovation in voice AI! 💡🎤 @shaheenawan @ruthbahr @stephaniewatts @micahboyer @robertbudinsky @yaelbensoussan #VoiceAI #DataCollection #ResearchInnovation #VoiceTechnology #Bridge2AI

Overfitting at its finest. Overfitting refers to a machine learning model that is too closely tailored to a particular dataset, leading to poor generalization on new data. A model that overfits is excellent at learning the details and noise of the training data but struggles to perform well on unseen data. 𝗪𝗵𝘆 𝗢𝘃𝗲𝗿𝗳𝗶𝘁𝘁𝗶𝗻𝗴 𝗛𝗮𝗽𝗽𝗲𝗻𝘀: 1. Complex Models: Overly complex algorithms can lead to overfitting by memorizing rather than learning patterns 2. Insufficient Data: A small training dataset can cause model to latch onto noise or minor patterns in the data. 3. Lack of Regularization: Without constraints, models can become too flexible and overfit the training data. 𝗣𝗿𝗲𝘃𝗲𝗻𝘁𝗶𝗻𝗴 𝗢𝘃𝗲𝗿𝗳𝗶𝘁𝘁𝗶𝗻𝗴: 1. Simplify the Model: Use a simpler model architecture that generalizes better. 2. More Data: Train the model on a larger dataset to ensure it learns robust patterns. 3. Cross-Validation: Validate the model on unseen data to ensure it performs well in general. Like this over-specialized doghouse, we must ensure our models don’t just fit perfectly to the training set but generalize well to new data. click "Follow" for more interesting insights and resources on AI/ML & Data Science: Arpit Singh • • #machinelearning #datascience #deeplearning #ai #ml

[English and Spanish below] 🇸🇰 SVK: ESLA Kongres (Európsky kongres logopédie a rečovej terapie) je jeden z top eventov, ktorý sa venuje poruchám reči a ich terapii. Tento rok sa koná v Belgicku a Scicake na ňom aktívne participuje. Spolu s našou poradkyňou na klinickú logopédiu, Neus Calaf, PhD, SLP, mám tú česť prezentovať výsledky nášho prieskumu v Španielsku. Zisťovali sme, ako sa v Španielsku diagnostikuje dysartria a dozvedeli sme sa zásadné informácie. Sledujte Scicake a dozviete sa viac! 🇬🇧 ENG: The ESLA Congress (European Congress of Speech and Language Therapy) is one of the top events addressing speech disorders and their therapy. This year, it is being held in Belgium, and Scicake is actively participating. Along with our Speech-Language Pathology Advisor, Neus Calaf, PhD, SLP, I have the honor of presenting the results of our survey in Spain. We explored the current state of dysarthria assessment in Spain and uncovered essential information. Follow Scicake to learn more! 🇪🇸 ES: El Congreso de ESLA (Congreso Europeo de Logopedia y Terapia del Lenguaje) es uno de los eventos más importantes sobre los trastornos del habla y su terapia. Este año se celebra en Bélgica, y Scicake participa activamente. Junto con nuestra asesora en logopedia clínica, Neus Calaf, PhD, SLP, tengo el honor de presentar los resultados de nuestra encuesta en España. Investigamos cómo se evalúa la disartria en España y descubrimos información clave. ¡Sigue a Scicake para saber más! #Scicake #ESLA2024 #SpeechDisorders #SpeechAndLanguageTherapy #dysarthria #disartria #dysartria #AI

With today's R&R we are back on the topic of speech... but thinking about how long-term medication use can affect the changes many people with #ParkinsonsDisease (pwPD) experience This came up on my timeline with a great engaging summary by one of the authors Jiri Mekyska One to read if you’re interested in 💬 Speech as a marker of change 🧠 Impact of medication use on Parkinson's ✍ Analysis of speech subt6ypes If you’re interested in this also check out the buzz around the upcoming International Congress of Parkinson’s Disease and Movement Disorders using the #MDS2024 on LinkedIn Link to the paper available in the comments 👇

Reinforcement Learning-Based Adaptive Classification for Medication State Monitoring in Parkinson’s Disease https://lnkd.in/eG8-4iSg Published by Shuqair M., Jimenez-Shahed J., Ghoraani B. #ParkinsonsDisease #ReinforcementLearning #MedicationMonitoring #MachineLearning #HealthcareAI

Digital biomarkers for non-motor symptoms in Parkinson’s disease: the state of the art https://lnkd.in/e6MJdavt Published by Janssen Daalen J.M., van den Bergh R., Prins E.M., Moghadam M.S.C., van den Heuvel R., et al. #ParkinsonsDisease #DigitalBiomarkers #NonMotorSymptoms #WearableTechnology #RemoteMonitoring #NeurodegenerativeDisease

Reducing the burden of care: Effect of art and dance therapy on Parkinson’s disease patients’ caregivers https://lnkd.in/es6-6SeB Published by Jagota P., Karmin S., Mashiach S., Federman D.J., Simhony M.F., et al. #ParkinsonsDisease #CaregiverSupport #ArtTherapy #DanceTherapy #MentalHealth