A Speech-Based Classification Model for Mild Cognitive Impairment Screening
Bibliographic record
Abstract
Mild Cognitive Impairment (MCI) is an early stage of cognitive decline that often goes undetected due to its overlap with normal aging symptoms. Early identification is crucial, as MCI can progress to dementia, a neurodegenerative condition that severely impairs quality of life and has become a growing global public health concern. While traditional cognitive screening tools, such as the Mini-Mental State Examination (MMSE) and Montreal Cognitive Assessment (MOCA), are commonly used, they suffer from subjectivity, cultural bias, and limited scalability, highlighting the need for objective, accessible alternatives. This study explored a non-invasive and scalable alternative to traditional MCI screening through speech-based methods. A speech-based classification model was built using an open-source database of audio recordings from MCI patients and healthy controls. To enhance detection performance, the model utilized language-independent acoustic features including Mel-Frequency Cepstral Coefficients (MFCC), jitter, shimmer, and fundamental frequency. Multiple machine learning models were trained and evaluated, including Random Forest, Support Vector Machine (SVM) and Logistic Regression, Convolutional Neural Network-Long Short-Term Memory (CNN-LSTM), and a lightweight Deep Neural Network (DNN). Random Forest performed the best among the baseline models with an accuracy of 83.5%. Between CNN-LSTM and DNN, DNN achieved the highest performance with an accuracy of 84%. These findings demonstrate the potential of universal speech features in supporting early detection of MCI through deep learning techniques while maintaining low computational cost. This approach underscores the potential of deep-learning-based speech biomarkers to enable non-invasive, scalable, and language-agnostic MCI screening, with implications for both clinical workflows and self-assessment.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.000 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".