MétaCan
Menu
Back to cohort
Record W4399773302 · doi:10.1101/2024.06.17.24308959

Frontotemporal Dementia Subtyping using Machine Learning, Multivariate Statistics, and Neuroimaging

2024· preprint· en· W4399773302 on OpenAlexafffund
Amelie Metz, Yashar Zeighami, Simon Ducharme, Sylvia Villeneuve, Mahsa Dadar

Bibliographic record

VenuemedRxiv · 2024
Typepreprint
Languageen
FieldNeuroscience
TopicBrain Tumor Detection and Classification
Canadian institutionsMontreal Neurological Institute and HospitalMcGill UniversityDouglas Mental Health University Institute
FundersFonds de Recherche du Québec - SantéCanadian Institutes of Health ResearchCompute CanadaNatural Sciences and Engineering Research Council of CanadaFondation Brain CanadaBiogenEli Lilly and CompanyAlnylam PharmaceuticalsNovo NordiskEisaiAlzheimer's Association
KeywordsSubtypingFrontotemporal dementiaNeuroimagingMultivariate statisticsDementiaMultivariate analysisPsychologyComputer scienceArtificial intelligenceCognitive psychologyMachine learningMedicineNeuroscienceInternal medicine

Abstract

fetched live from OpenAlex

Abstract Frontotemporal Dementia (FTD) is a prevalent form of early-onset dementia characterized by progressive neurodegeneration. It encompasses a group of heterogeneous disorders, including behavioral variant frontotemporal dementia (bvFTD), nonfluent variant primary progressive aphasia (nfvPPA), and semantic variant primary progressive aphasia (svPPA). Due to disease heterogeneity and overlapping symptoms, diagnosis of FTD and its subtypes still poses a challenge. Magnetic-resonance imaging (MRI) is commonly used to support the diagnosis of FTD. Using machine learning and multivariate statistics, we tested whether brain atrophy patterns are associated with severity of cognitive impairment, whether this relationship differs between the phenotypic subtypes, and whether we could use these brain patterns to classify patients according to their FTD variant. A total of 136 patients (70 bvFTD, 36 svPPA, 30 nfvPPA) from the frontotemporal lobar degeneration neuroimaging initiative (FTLDNI) database underwent brain MRI and clinical and neuropsychological examination. Deformation-based morphometry (DBM), which offers increased sensitivity to subtle local differences in structural image contrasts was used to estimate regional cortical and subcortical atrophy. Atlas-based associations between DBM values and performance across different cognitive tests were assessed using partial least squares (PLS). We then applied linear regression models to discern the group differences regarding the relationship between atrophy and cognitive decline in the three FTD phenotypes. Lastly, we assessed whether the combination of neural and behavioral patterns in the latent variables identified in the PLS analysis could be used as features in a machine-learning model to predict FTD subtypes in patients. Results revealed four significant latent variables that combined accounted for 86% of the shared covariance between cognitive and brain atrophy measures. PLS-based atrophy and behavioral patterns predicted the FTD phenotypes with a cross-validated accuracy of 89.12%, with high specificity (91.46-97.15%) and sensitivity (84.19-93.56%). When using only MRI measures and two behavioral tests in the PLS and classification algorithm, ensuring clinical feasibility, our model was similarly precise (83.62%, specificity 86.38-93.51%, sensitivity 76.17-87.50%). Here, including only atrophy or behavior patterns in the analysis led to prediction accuracies of 69.76% and 76.38%, respectively, highlighting the increased value of combining MRI and clinical measures in subtype classification. We demonstrate that the combination of brain atrophy and clinical characteristics, and multivariate statistical methods can serve as an imaging biomarker for early disease phenotyping in FTD, whereby inclusion of DBM measures adds to the classification precision in the absence of extensive clinical testing.

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 distilled prediction

Teacher imitation

Not calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.

metaresearch head score (Codex)0.000
metaresearch head score (Gemma)0.001
Version: codex-gemma-dda1882f352aValidation status: machine_predicted_unvalidated
Candidate categoriesMeta-epidemiology (narrow)
Consensus categoriesnone
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Simulation or modeling · Consensus signal: none
GenreCandidate signal: Empirical · Consensus signal: Empirical
Teacher disagreement score0.947
Threshold uncertainty score1.000

Codex and Gemma teacher scores by category

CategoryCodexGemma
Metaresearch0.0000.001
Meta-epidemiology (narrow)0.0000.000
Meta-epidemiology (broad)0.0000.000
Bibliometrics0.0000.000
Science and technology studies0.0000.000
Scholarly communication0.0000.000
Open science0.0000.001
Research integrity0.0000.001
Insufficient payload (model declined to judge)0.0000.000

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.

Opus teacher head0.063
GPT teacher head0.308
Teacher spread0.245 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it

Classification

machine, unvalidated

Machine predicted; a candidate call from one teacher head, not a consensus.

Study designSimulation or modeling
Domainnot available
GenreEmpirical

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".

Quick stats

Citations4
Published2024
Admission routes2
Has abstractyes

Explore more

Same venuemedRxivSame topicBrain Tumor Detection and ClassificationFrench-language works237,207