Searcharxiv⌕ Search

arXiv subjects

Cristina Martin

Publications and source records attributed to Cristina Martin.

3 recordsLinked to original sources

Addressing Bias Early in the AI Lifecycle: Social Determinants of Health in Functional and Cognitive Assessment Labels of Nursing Home Residents

Fairness in clinical artificial intelligence (AI) is usually evaluated after model training, yet bias can enter earlier in the AI lifecycle, during data collection and labelling. When routine clinical assessments serve as target variables, systematic differences in these labels across sociodemographic groups are carried into model development, validation and updating. This study addresses this early lifecycle stage by auditing labels from three standardised instruments used in residential care: the Barthel Index (BI), the Mini-Mental State Examination (MMSE) and the Global Deterioration Scale (GDS). Using electronic health records of 2,407 residents from 12 nursing homes in the Basque Country, Spain (2011-2021), we assessed four social determinants of health (SDoH) - sex, educational attainment, lifetime work activity and marital status - with two group fairness metrics, Disparate Impact (DI) and Statistical Parity Difference (SPD), computed for each outcome category with bootstrap confidence intervals. Disparities were found for all four attributes and across the three instruments: men, residents with intermediate education and never-partnered residents were more frequently classified in categories indicating preserved function or cognition. These disparities may reflect genuine differences in health status, measurement properties of the instruments, assessment practices or selective admission to residential care, and cannot be attributed to a single cause with the present design. Regardless of their origin, any model trained on these labels would inherit them, and they could persist through later lifecycle stages unless explicitly examined. These findings support pre-training label auditing as the first step of a continuous fairness assessment spanning model development, validation, deployment and updating.

cs.LO↗

Biotransformation and biological impact of graphene and graphene oxide during simulated oral ingestion

Graphene is an innovative nanomaterial, made of two-dimensional honeycomb-like carbon lattice, with potential in many different applications. Studying the behaviour of graphene-related materials (GRMs) in biological systems is, therefore, crucial to assess possible side effects. In this work is studied the biotransformation and biological impact of few-layer pristine graphene (FLG) and graphene oxide (GOX), following ingestion as exposure route. To mimic FLG and GOX ingestion, an in vitro digestion assay based on a standardized operating procedure (SOP) is applied. The assay simulates the human ingestion of GRMs during their dynamic passage through the different environments of gastrointestinal (GI) tract (salivary, gastric, intestinal). Physical-chemical changes of GRMs during the digestion process are assessed through a detailed Raman spectroscopy characterization. Moreover, the effect of chronic exposure to digested GRMs on integrity and functionality of an in vitro model of intestinal barrier is also determined according to a second SOP. Our results showed that the FLG and GOX nanoflakes aggregates after experiencing the passage through the different environments of GI tract, with evident doping effects, due to the interaction of GRMs with digestive juice components and the strong pH variations. Interestingly, neither structural changes nor degradation of the nanomaterials is observed, suggesting that GRMs are bio-persistent nanomaterials when administered by oral route. Chronic exposure to digested GRMs does not affect intestinal barrier integrity and is not associated to inflammation and cytotoxicity, though possible long-term adverse effects cannot be ruled out, due to the observed bio-durability of the materials.

cond-mat.soft↗

Soil Moisture Monitorization Using GNSS Reflected Signals

The use of GNSS signals as a source of opportunity for remote sensing applications, GNSS-R, has been a research area of interest for more than a decade. One of the possible applications of this technique is soil moisture monitoring. The retrieval of soil moisture with GNSS-R systems is based on the variability of the ground dielectric properties associated to soil moisture. Higher concentrations of water in the soil yield a higher dielectric constant and reflectivity, which incurs in signals that reflect from the Earth surface with higher peak power. Previous investigations have demonstrated the capability of GPS bistatic scatterometers to obtain high enough signal to noise ratios in order to sense small changes in surface reflectivity. Furthermore, these systems present some advantages with respect to others currently used to retrieve soil moisture. Upcoming satellite navigation systems, such as the European Galileo, will represent an excellent source of opportunity for soil moisture remote sensing for various reasons. First, the existence of pilot signals will provide the possibility to extend coherent integration times, which will contribute to the increase of received signals SNR. In addition, the availability of Galileo L1 and L5 signals will allow the multi-spectral analysis of the reflected signals and the development of inversion models which will be able to account more precisely for adverse effects, such as surface roughness and vegetation canopy. In this paper we present some of the recent theoretical work and experiments carried out at Starlab focusing on the development of dedicated Soil Moisture GNSS-R systems.

physics.geo-ph↗