Dataset Information

Clustering: how much bias do we need?

ABSTRACT: Scientific investigations in medicine and beyond increasingly require observations to be described by more features than can be simultaneously visualized. Simply reducing the dimensionality by projections destroys essential relationships in the data. Similarly, traditional clustering algorithms introduce data bias that prevents detection of natural structures expected from generic nonlinear processes. We examine how these problems can best be addressed, where in particular we focus on two recent clustering approaches, Phenograph and Hebbian learning clustering, applied to synthetic and natural data examples. Our results reveal that already for very basic questions, minimizing clustering bias is essential, but that results can benefit further from biased post-processing.This article is part of the themed issue 'Mathematical methods in medicine: neuroscience, cardiology and pathology'.

SUBMITTER: Lorimer T

PROVIDER: S-EPMC5434083 | biostudies-literature | 2017 Jun

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Clustering: how much bias do we need?

Lorimer Tom T Held Jenny J Stoop Ruedi R

Philosophical transactions. Series A, Mathematical, physical, and engineering sciences 20170601 2096

Scientific investigations in medicine and beyond increasingly require observations to be described by more features than can be simultaneously visualized. Simply reducing the dimensionality by projections destroys essential relationships in the data. Similarly, traditional clustering algorithms introduce data bias that prevents detection of natural structures expected from generic nonlinear processes. We examine how these problems can best be addressed, where in particular we focus on two recent ...[more]

PMID: 28507238

Similar Datasets

Project description:BackgroundTo investigate age-related severity, patterns of retinal structural damage, and functional visual recovery in pediatric and adult cohorts of myelin oligodendrocyte glycoprotein antibody-associated disease (MOGAD) optic neuritis (ON).MethodsAll MOGAD patients from the 5 participating centers were included. Patients with initial manifestation <18 years were included in the pediatric (MOGADped) cohort and patients with ≥18 years in the adult (MOGADadult) cohort. For patients with MOGAD ON, examinations at least ≥6 months after ON onset were included in the analyses. Using spectral domain optical coherence tomography (SD-OCT), we acquired peripapillary retinal nerve fiber layer thickness (pRNFL) and volumes of combined ganglion cell and inner plexiform layer (GCIPL). High- and 2.5% low-contrast visual acuity (HCVA, LCVA) and visual-evoked potentials (VEP) were obtained.ResultsTwenty MOGADped (10.3±3.7 years, 30 MOGAD ON eyes) and 39 MOGADadult (34.9±11.6 years, 42 MOGAD ON eyes) patients were included. The average number of ON episodes per ON eye was similar in both groups (1.8±1.3 and 2.0±1.7). In both pediatric and adult MOGAD, ON led to pronounced neuroaxonal retinal atrophy (pRNFL: 63.1±18.7 and 64.3±22.9 μm; GCIPL: 0.42±0.09 and 0.44±0.13 mm3, respectively) and moderate delay of the VEP latencies (117.9±10.7 and 118.0±14.5 ms). In contrast, visual acuity was substantially better in children (HCVA: 51.4±9.3 vs. 35.0±20.6 raw letters, p=0.001; LCVA: 22.8±14.6 vs. 13.5±16.4, p=0.028). Complete visual recovery (HCVA-logMAR 0.0) occurred in 73.3% of MOGADped and 31% MOGADadults ON eyes, while 3.3% and 31% demonstrated moderate to severe (logMAR > 0.5) visual impairment. Independent of retinal atrophy, age at ON onset significantly correlated with visual outcome.ConclusionPediatric MOGAD ON showed better visual recovery than adult MOGAD ON despite profound and almost identical neuroaxonal retinal atrophy. Age-related cortical neuroplasticity may account for the substantial discrepancy between structural changes and functional outcomes.

Dataset Information

Clustering: how much bias do we need?

Publications

Clustering: how much bias do we need?

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure