Browse Articles

Behavioral determinants of continued use of Islamic FinTech: The moderating role of service and system quality

The purpose of this study is to investigate the determinants of continued use of Islamic FinTech services through an examination of how useful Islamic FinTech services appear, what others think about the user’s use of Islamic FinTech services, how confident the user feels when using these services, and how much the user trusts the service provider in relation to the user’s actual use. This study will examine the relationship between the user’s actual use and their behavioral intention to continue to use the service. Additionally, this study will examine the role that both the service quality and system quality play as moderators for post adoption behaviors. In order to complete this study, a cross sectional survey was used to gather data from 232 active users of Islamic FinTech services in Jordan. A series of previously validated likert scale items were used to measure the variables used in this study. PLS-SEM with SmartPLS 4 software was utilized to test both direct and moderator effects in the proposed research model. Bootstrapping techniques were applied in order to determine if the structural paths in the model were significantly different from zero at a 95% confidence level. The results indicated that the proposed model accounted for 63.4% of the variance in actual use (R² = .634). The findings show that all four constructs; perceived usefulness, subjective norms, self-efficacy and perceived trust, are positively related to the user’s actual use of Islamic FinTech services. Also, there is a positive association between actual use and behavioral intentions to continue use of Islamic FinTech services. System quality has a moderating effect on the relationship between actual use and continuance intention while service quality does not have a statistically significant moderating effect. This study is one of few studies examining post adoption behavior in the context of Islamic FinTech services within Jordan. The study combines multiple theoretical frameworks including cognitive theory, social influence theory, and trust theory and incorporates both service quality and system quality into a singular framework to examine continuance behavior of Islamic FinTech usage.

Distinct Response Selectivity Changes in the Primary Visual and Parietal Cortex during Visual Discrimination Learning

When animals learn the behavioral relevance of sensory features, response selectivity in primary sensory areas increases for those features. However, the effect of learning on neuronal activity in higher-level areas associated with decision-making in the parietal cortex, compared with primary sensory cortex, is not yet known. We used two-photon calcium imaging to determine how learning modifies neural representations in the primary visual cortex (V1) and posterior parietal cortex (PPC) in a visual go/no-go orientation discrimination task in male and female mice. We found that behavior improvements after learning were associated with increased neuronal selectivity in both V1 and PPC. The increased selectivity in PPC was mainly driven by neurons preferring the rewarded go stimulus, while neurons in V1 increased their selectivity for both the rewarded go and unrewarded no-go stimulus. Furthermore, feature preference was robust in V1 neurons but reorganized in PPC after learning. Finally, feature preference after learning was preserved across contexts in V1 but not in PPC, where many neurons switched their preference to the rewarded feature during active task engagement. Our results demonstrate that learning a visually guided discrimination task increased information about relevant sensory features through distinct changes in the bottom and top levels of the visual cortical hierarchy. Visual cortex neurons encode visual features with increased reliability but with preserved feature preferences after learning, while parietal neurons reorganize their feature preferences in a task-dependent manner.

Towards conversational artificial intelligence for disease management

Abstract Although large language models have shown promise in diagnostic dialogue 1 , their capabilities for effective management reasoning, including disease progression, therapeutic response and safe medication prescription, have remained underexplored. We have advanced the previously demonstrated diagnostic capabilities of the Articulate Medical Intelligence Explorer (AMIE) 1–3 using a new large-language-model-based agentic system optimized for multivisit clinical management and dialogue. To ground the reasoning of AMIE in authoritative clinical knowledge, we leveraged the long-context capabilities of Gemini 4 , combining in-context retrieval with structured reasoning to align its output with up-to-date clinical practice guidelines and drug formularies. In a randomized, blinded virtual Objective Structured Clinical Examination study, AMIE was compared to 21 primary care physicians (PCPs) across 100 multivisit case scenarios designed to reflect the guidance of the UK National Institute for Health and Care Excellence and BMJ Best Practice guidelines. AMIE was non-inferior to PCPs in management reasoning, as assessed by specialists, and scored better both with respect to preciseness of treatment and investigation, and in terms of its alignment with and grounding in clinical guidelines. To benchmark medication reasoning, we developed RxQA, a multiple-choice question benchmark that was derived from two national drug formularies (from the USA and UK) and validated by board-certified pharmacists. Although AMIE and PCPs both benefited from the ability to access external drug information, AMIE outperformed PCPs on higher-difficulty questions. Although further research will be needed before real-world translation of AMIE, its strong performance across evaluations marks a significant step towards use of conversational artificial intelligence as a tool in disease management.

Chirality Transfer, Memory and Sensing Activated by a Supramolecular Chiral Auxiliary Approach in Nanostructured, Tautomerically Prochiral Triptycene-Fused Benzimidazoles

Abstract We report the synthesis and characterization of stimuli responsive, adaptive organic chiral nanoparticles assembled using a “stereochemically fluid”, triptycene-fused benzimidazole, which is chiroptically activated by the noncovalent interaction with enantiopure tartaric acid, acting as a “supramolecular chiral auxiliary”. Under appropriate experimental conditions, the formation of chiral supramolecular aggregates exhibits remarkable chiroptical properties (electronic circular dichroism, ECD, and circularly polarized luminescence, CPL) with the possibility of addressing their controlled manipulation, reversion between the states, and chirality memory properties. Furthermore, the nanostructured system is capable of selective sensing of the Cu2+ ion through the modulation of its chiroptical properties. In this manner, we provide an unprecedented way to introduce chirality onto the triptycene skeleton to assemble well-defined chiral nanoparticles with dimensions of several hundred nanometers and to reversibly store chiral information and activate chiroptical sensing properties.

Genetic diversity, population structure, and combined detection of selection signatures in Iranian versus Afghan Baluchi sheep

Selection to increase the frequency of useful mutations has left marks on animal genomes, genetic diversity, and population structure within populations. The study and investigation of these genomic regions can lead to the identification of genes related to economic traits or competence and adaptability. This study aimed to recognize genetic diversity, population structure, and selection signatures in Iranian (IB) and Afghan (AB) Baluchi sheep populations. In this study, 86 Iranian Baluchi and 15 Afghan Baluchi sheep were genotyped using Illumine Ovine SNP50K Beadchip arrays. Note that the sample size imbalance (IB n = 86 vs. AB n = 15) may reduce statistical power and potentially bias population structure and selection scan results. Additionally, use of the Ovine 50K array may introduce ascertainment bias; analyses were based on 38,193 shared SNPs, potentially missing population-specific variants. Generally, moderate genetic diversity was observed in both the Afghan Baluchi (AB) and Iranian Baluchi (IB) sheep populations, using various assessment methods. However, the IB population showed the lowest level of genetic diversity and the highest rate of linkage disequilibrium decay, despite having a better effective population size in recent generations. The ADMIXTURE analysis indicated that the optimal number of genetic clusters was K = 2, which was determined based on the lowest cross-entropy error of 0.603 observed during cross-validation. At K = 2 and NJ tree analysis, a clear genetic distinction between the AB and IB populations was evident. Additionally, the IB population demonstrated significant genetic uniformity when compared to the AB population in terms of genetic distance. Also, F ST and XP-EHH were used to identify selection signatures. Some putative candidate genes for F ST, including HDAC9 , CSMD3 , DAB1 , FGF12 , and PCDH9 were associated with important economic traits such as body weight, hot carcass weight, muscle weight in carcass, reproductive seasonality, and carcass fat percentage, respectively. Also, XP-EHH putative candidate genes were KCNIP4 , FGF11 , CNTROB , and ROBO2 in AB population, which were related to body weight, hot carcass weight, milk yield, and muscle weight in carcass. Moreover, XP-EHH putative candidate genes in IB population were GRIK3 , NCOA1 , and FGD3, that related to muscle weight in carcass, staple length, and milk fat percentage. Selection signals were identified using top 1% F ST thresholds and XP-EHH without genome-wide multiple-testing correction; results require experimental validation. We observed very similar outcomes in terms of similar signatures related to economic traits in both F ST and XP-EHH methods, indicating the robustness of analysis in this study. It can be concluded that selection has made a major distinction between Afghan and Iranian Baluchi sheep populations for reproduction, milk production, and growth traits. These could be due to the managed breeding programme in Iranian Baluchi sheep. Utilizing validated QTLs as described in this study could be applied to reveal the direction of breeding plans in livestock species.

Regulation of Schwann Cell–Axon Interactions in Nerve Development: The Role of Slit2 in Axonal Sorting

Radial sorting of axons is a critical process in nerve development, ensuring proper segregation of axons to form myelinated and unmyelinated Schwann cell–axon units. This process is regulated by signals mediating communication between Schwann cells and the extracellular matrix, with laminin-211 as a key component. However, the molecular signals involved in directing Schwann cell–axon interactions are less understood, highlighting the need to identify additional molecules that mediate axon recognition and segregation. Gaining a deeper understanding of these mechanisms may shed light on the pathogenesis of genetic neuropathies. In this study, we utilized a mouse model of either sex with defects in axonal sorting, resulting from the conditional inactivation of the COP9 signalosome component Csn5 ( Jab1 ) in Schwann cells. Transcriptome analysis was performed to identify adhesion molecules dysregulated during nerve development. Notably, we discovered that the repulsive molecule Slit2 was significantly overexpressed in Jab1-KO nerves and was particularly abundant in axon bundles with improper sorting. We demonstrated that while Slit2 is highly expressed in embryonic nerves, its expression must be precisely regulated in mature Schwann cells. Gain- and loss-of-function mutants for Slit2 further confirmed the role of Slit2 in nerve development. Transgenic mice overexpressing Slit2 displayed defects in radial sorting and hypomyelination, while Slit2 loss led to hypermyelination and misalignment of unmyelinated axons in Remak bundles. Additionally, Slit2 dysregulation interfered with nerve regeneration following cut injury. Our findings suggest that Slit2 plays a significant role in multiple stages of nerve development and some aspects of nerve regeneration.

How the zebrafish brain weaves recent experiences into future decisions

Association between results of component-resolved diagnostics and basophil activation in Hymenoptera venom allergy: A registry-based cross-sectional study in adults

Background Allergy to Hymenoptera venom is one of the most frequent causes of anaphylaxis in adults. Conventional diagnostic approaches, including skin testing and measurement of allergen-specific IgE (sIgE) to venom extracts, do not always allow for precise identification of the culprit venom. This remains particularly challenging in patients with double-positive or inconclusive conventional test results, which may complicate the decision-making process regarding qualification for appropriate treatment. Component-resolved diagnostics (CRD) and the basophil activation test (BAT) may provide complementary information in such cases and aid in qualification for subcutaneous immunotherapy (SCIT). Methods In this retrospective registry-based cross-sectional study, 154 adults who had been evaluated for Hymenoptera venom allergy at the Military Institute of Medicine (Warsaw, Poland) between December 2023 and May 2025 were included. Patients were divided into three groups: not qualified for SCIT (n = 27), qualified for bee venom immunotherapy (n = 32), and qualified for wasp venom immunotherapy (n = 95). Serum sIgE to venom extracts and certain components (rApi m 1, m 2, m 3, m 5, m 10; rVes v 1, v 5) were measured. BAT was performed by flow cytometry assessing CD63 expression. Results Moderate correlations were found between BAT results and sIgE to rApi m 1 (rho = 0.495; p < 0.001) and rVes v 5 (rho = 0.456; p < 0.001). No correlation was observed for rVes v 1, and negative correlations were noted between rApi m 1 and rVes v 5 in heterologous BAT responses. No statistically significant association was observed between the severity of previous sting reactions and BAT or sIgE parameters. However, due to limited subgroup sizes, these analyses were underpowered and the results should be interpreted as inconclusive. Conclusions Among the examined variables, the major components Api m 1 and Ves v 5 were the most strongly correlated with specific basophil activation in vitro. The combined use of CRD and BAT may support clinical assessment and facilitate decisions regarding immunotherapy qualification. Further prospective studies are warranted to assess the prognostic and clinical relevance of our findings.

Implementing a patient-oriented pole walking intervention in retirement homes: A non-randomized feasibility trial

Objectives To evaluate the feasibility and safety of implementing a patient-oriented pole walking (PW) intervention in retirement home settings and preliminary changes in outcome measures related to physical function and other fall- and fracture-related risk factors to inform a future randomized controlled trial (RCT). Methods This single-arm, non-randomized feasibility trial implemented a patient-oriented PW intervention across four retirement homes in Saskatoon, Saskatchewan, Canada. During Summer 2022, we assessed 24 residents for eligibility, of which 19 consented and 17 received the intervention. The intervention was offered as supervised group sessions (20–60 minutes) three times per week for 12 weeks. Each session consisted of posture and balance warm-up, PW, muscle strengthening, and stretching. The primary outcome measure was feasibility as assessed by consent, recruitment, retention, and adherence rates as well as by intervention acceptability, appropriateness, and feasibility scores. The secondary outcome measures included safety (evaluated by recorded adverse events) and preliminary 12-week changes in physical function and other fall- and fracture-related risk factors (examined with paired-samples t-tests or repeated measures analysis of covariance models). Results Fifteen participants (mean age 85.2 years; 93% female) completed the study. The consent, recruitment, retention, and mean adherence rates were 79%, 2.7 participants/site/month, 88%, and 90%, respectively. The mean participant- and instructor-reported scores for intervention acceptability, appropriateness, and feasibility were all > 4.0 (out of 5). There were no recorded intervention-related serious adverse events. Participants improved their functional balance/mobility (timed “up & go” test: −1.6 seconds; 95% CI: −2.7 to −0.4), lower-body strength (30-second chair stand test: 2.4 repetitions; 1.2 to 3.5), 36-item short-form survey physical functioning score (12.9; 3.7 to 22.2), and forearm muscle area (67.7 mm 2 ; 12.9 to 122.6) over 12 weeks. Conclusions It was feasible and safe to implement our patient-oriented PW intervention in retirement homes. Findings will inform our future RCT in these settings. Trial registration ClinicalTrials.gov NCT05388227 .

Brain Representations of Natural Sound Statistics

Natural sound textures (e.g., rain, crackling fire) are perceptually defined by time-averaged summary statistics. While previous studies have examined neural responses to natural sounds, little is known regarding the neural processing of the statistics underlying these sounds. To study neuronal correlates of these statistics, we measured brain responses to synthetic sound textures in which statistical structure was systematically varied while preserving the texture category. Using two fMRI experiments (males and females), we examined neural responses along the ascending auditory pathway, within auditory cortex and medial temporal lobe (MTL) regions implicated in pattern analysis. In Experiment 1, we parametrically varied the full set of texture statistics, creating sounds with different levels of naturalness. In Experiment 2, we selectively manipulated high-level statistics (cochlear skewness and kurtosis, cochlear and modulation correlations) while holding low-level statistics (cochlear mean and modulation power) constant. Increasing texture naturalness produced graded increases in BOLD responses across bilateral primary and nonprimary auditory cortex in both experiments, although overall responses were weaker in Experiment 2. This reduction suggests that low-level statistics contribute substantially to response magnitude, although higher-order statistics are sufficient to elicit graded responses. We also observed modulation in MTL regions, including entorhinal cortex, in Experiment 1. Moreover, functional connectivity between hippocampus and auditory cortex increased for more degraded (less natural) textures, suggesting a modulatory rather than representational role for MTL in texture processing. Together, these findings show that sensitivity to texture statistics is distributed across the auditory cortex and highlight MTL–auditory interactions when texture structure is ambiguous.

Impact of the FTO rs9939609 risk allele on subcutaneous adipose tissue fatty acid composition in adults with obesity class 2 and 3

The FTO rs9939609 risk allele is linked to risk of obesity. Whether causality involves fatty acid (FA) metabolism remains to be fully investigated in adults with obesity. We tested for associations of the risk allele with the FA composition of android and gynoid subcutaneous adipose tissue. We recruited 95 participants with obesity class 2 and 3 and without diabetes, median BMI 42.8 (25 th , 75 th percentiles: 39.5, 46.5) kg/m 2 . Participants carried no (TT, n = 33), one (AT, n = 31), or two (AA, n = 31) copies of the FTO risk allele. Biopsies were obtained by aspiration and total FA composition determined by gas chromatography-mass spectrometry (GC-MS). In the cohort overall, there were no significant genotype associations with any single FA. In males with the TT allele, mass of oleic acid (18:1n-9) in the gynoid depot was higher compared with the AT allele when corrected for depot size. We interpret these findings with caution due to the small numbers of males with the TT genotype. Disregarding genotype, in the cohort overall, proportions of saturated FAs were higher, and proportions of monounsaturated FAs lower in android versus gynoid adipose tissue, confirming previous studies. We found previously unreported sex-related differences in FA composition (weight %) and content (weight % corrected for depot mass). Our findings on FTO genotype are generally negative; observations to the contrary require confirmation as does the non-genetic and novel results on sex-related differences.

Mapping the neuronal building blocks of human language with language models

Adverse events following HPV vaccine reported to the Vaccine Adverse Event Reporting System

Background Vaccination against high-risk HPV types is a key preventive measure. However, concerns regarding vaccine safety may hinder vaccination efforts. This study aims to evaluate adverse events (AEs) reported in the Vaccine Adverse Event Reporting System (VAERS) following HPV vaccination from 2006 to 2024, providing insights into its safety profile. Methods We analyzed VAERS data, a spontaneous reporting system containing de-identified AE reports. Four disproportionality analyses (ROR, PRR, BCPNN, and MGPS) were applied, and adverse event signals were defined only when the positive criteria were simultaneously met across all four methods. Statistical analyses were performed using R software and Microsoft Excel, with a significance threshold of p < 0.05. Results A total of 77,909 HPV vaccine-related AE reports were analyzed, with 68.4% involving females and 48.7% affecting individuals under 18. Serious AEs accounted for 11,659 reports, with headache and fatigue being the most common. Syncope was the most frequent signal, while postural orthostatic tachycardia syndrome (POTS) exhibited the strongest signal strength. Approximately 90% of AEs occurred within 30 days post-vaccination. Among vaccine types, HPV4 had the highest number of reports, and intramuscular injection was the most common administration route. Conclusion This study offers an updated pharmacovigilance assessment of adverse events reported following HPV vaccination, highlighting reported patterns and statistical signals that may warrant further investigation.

Depression, anxiety symptoms, and association with household characteristics in adolescent boys and girls from Matiari District, Pakistan: A community-based cross-sectional study

Introduction Pakistan has one of the world’s largest adolescent populations, yet evidence on the prevalence and correlates of depressive and anxiety symptoms in adolescents remains limited, particularly in rural settings. Objective This study aimed to estimate the prevalence of depressive and anxiety symptoms and examine their associations with household characteristics in a community-based sample of adolescents from the predominantly rural district of Matiari, Pakistan. Methods We examined cross-sectional data from 718 girls (9.0–14.9 years) and 678 boys (10.0–15.9 years) participating in the Nash-wo-Numa Study. Trained psychologists administered the Sindhi versions of the Short Mood and Feelings Questionnaire and the Screen for Child Anxiety Related Emotional Disorders to assess adolescents’ depressive and anxiety symptoms. Prevalence estimates and 95% confidence intervals were derived based on validated cut-off scores. Household correlates of depressive and anxiety symptoms were examined in multivariable negative binomial regression models. Results Approximately 8% of boys and 10% of girls exhibited clinically-significant depressive symptoms. The prevalence of clinically-significant anxiety symptoms ranged from 6% in boys and 8% in girls for generalized anxiety to 24% in boys and 39% in girls for separation anxiety symptoms. Girls experienced more depressive symptoms, panic/somatic and generalized anxiety symptoms than boys at age 12, more separation anxiety symptoms from age 11 onward, and more social anxiety symptoms from age 12 onward. In both sexes, depressive and anxiety symptoms were higher among adolescents exposed to intimate partner violence against their mothers and to moderate‑to‑severe food insecurity, and were lower among those with a homemaker mother. Among girls, maternal mental well‑being attenuated the association between food insecurity and depressive symptoms. Conclusion Depressive and anxiety symptoms are common among adolescents living in Matiari. Adolescents exposed to intimate partner violence against their mother, moderate-to-severe food insecurity, and poor maternal mental health may be at increased risk of depression and anxiety in predominantly rural Pakistan and may benefit from targeted prevention and intervention strategies.

Structure of the pre-initiation complex explains CMGE biogenesis

Abstract When cells enter S phase, bidirectional DNA replication is initiated through the kinase-regulated recruitment of three activators (Cdc45, GINS and Pol ε) to a duplex-DNA-loaded double hexamer of minichromosome maintenance (MCM) ATPases. Together, these proteins form two CMGE helicases that establish divergent replication forks as they become separated 1 . Here, to gain an understanding of CMGE biogenesis, we reconstituted the pre-initiation complex with purified yeast proteins. The cryo-electron-microscopy structure shows a set of firing factors caught in the act of assembling two symmetrical CMGEs. We show how stepwise complex formation reshapes MCM in preparation for DNA opening, and we explain how ATP promotes firing-factor ejection and CMGE maturation. We find that although Sld2 facilitates the recruitment of GINS to MCM, as expected, it also aids the efficient separation of the CMGE dimer, and is essential for the ejection of the lagging strand from MCM. These findings have direct implications for our understanding of the metazoan Sld2 orthologue, RECQL4, and point to a replication-fork establishment mechanism that is conserved across eukaryotes.

Scalable Topochemical Synthesis of Black Phosphorene Nanoribbons

Expression of concern: Engineering Pseudomonas protegens Pf-5 for nitrogen fixation and its application to improve plant growth under nitrogen-deficient conditions

Targeting the HMGB1–TLR4 Axis Alleviates Neuropathic Pain-Associated Cognitive Deficits

Cognitive deficits associated with chronic pain pose a significant burden on a patient's quality of life. Emerging evidence indicates that Toll-like receptor 4 (TLR4), a pattern recognition receptor implicated in neuroinflammatory signaling, can disrupt synaptic plasticity and memory processes. However, the specific involvement of TLR4 in the development of neuropathic pain-related cognitive deficits has not been fully elucidated. In this investigation, we observed an upregulation of TLR4 expression within hippocampal neurons in male mice subjected to chronic constriction injury (CCI) relative to the sham group. Notably, in separate experimental cohorts, TLR4 knock-out and neuron-specific TLR4 knockdown mice exhibited improved cognitive function compared with wild-type controls, alongside attenuated neuroinflammatory responses, reduced neuronal apoptosis, and enhanced preservation of hippocampal neuroplasticity. Concurrently, elevated concentrations of high-mobility group box 1 (HMGB1), a damage-associated molecular pattern molecule, were detected in the sciatic nerve, serum, and hippocampal tissues following CCI. Furthermore, increased colocalization of HMGB1 with TLR4 was evident in the hippocampus. Exogenous administration of HMGB1 augmented HMGB1 and TLR4 levels in the hippocampus and worsened memory functions that depend on hippocampal integrity. Conversely, inhibition of HMGB1 with glycyrrhizin, which subsequently attenuates TLR4 activation, ameliorated cognitive impairments induced by CCI. Collectively, these results support a model in which HMGB1, elevated during chronic neuropathic pain, contributes to cognitive deficits via a TLR4-dependent mechanism, triggering downstream inflammatory and apoptotic cascades and impairing synaptic plasticity.

Tailoring Local–Global Structures via Hot Deformation for High-Performance BiSbSe3 Thermoelectrics

Abstract Enhancing carrier concentration (n) is widely regarded as a core strategy for advancing high-performance thermoelectric (TE) materials. However, this approach is often limited by a concomitant decline in carrier mobility (μ). To surmount this trade-off, a synergistic integration combining composite engineering and hot deformation processing was employed to synergistically optimize both n and μ in BiSbSe3. Microstructural analysis reveals that this dual processing route drives a local–global structural evolution, involving texture formation, dynamic recrystallization, precipitation of Cu-rich secondary phases (CuSbSe2), incorporation of interstitial Cu atoms, and enhanced short-range ordering. As a result, the optimized n and tailored carrier transport pathways lead to reproducible and substantially enhanced electrical conductivity and power factor. Meanwhile, interstitial atoms, dislocations, subgrain boundaries, and heterogeneous interfaces collectively create a multiscale phonon scattering network, effectively reducing lattice thermal conductivity. Consequently, the peak ZT value along the out-of-plane direction is dramatically enhanced from ∼0.06 for pristine BiSbSe3 to ∼1.3 at 723 K for the BiSbSe3 + 2 mol % CuI + 1.8 mol % Cu sample subjected to single-pass hot deformation. This peak ZT value surpasses the highest reported value at the same temperature, with the Vickers hardness of the modified sample concurrently improved. This work elucidates the micromechanisms through which hot deformation synergistically regulates TE properties via tailoring of local–global structural modifications, laying a solid foundation for future commercialization.

Automated classification of natural habitats using ground-level imagery

Accurate classification of terrestrial habitats is critical for biodiversity conservation, ecological monitoring, and land use planning. Several habitat classification schemes are in use, typically based on analysis of satellite imagery and validation by field ecologists. Here, a methodology is presented for classification of habitats based solely on ground-level imagery (photographs), offering improved validation and enhanced ability to classify habitats at scale (e.g., using imagery from citizen science). In collaboration with Natural England, a public sector organisation with responsibility for nature/biodiversity conservation in England, this study develops a classification system that applies deep learning to ground-level habitat photographs, categorising each image into one of 16 distinct classes following the established ‘Living England’ framework. Images were pre-processed using resizing, normalisation, and augmentation techniques, while resampling was used to balance classes in the training data and enhance model robustness. A custom deep learning classifier based on the DeepLabV3-ResNet101 architecture was developed and fine-tuned to assign a habitat class label to ground-level photographs. Using five-fold cross-validation, the model demonstrated strong overall performance across 16 habitat classes, with accuracy and F1-scores varying between classes. This approach supports robust, scalable habitat classification based on balanced and well-prepared training data. Across all folds, the model achieved a mean F1-score of 0.63, with some habitat classes such as Bare Sand (BS) and Coniferous Woodland (CW) reaching values above 0.87. High performance was achieved for visually distinct habitats and lower performance for visually mixed or ambiguous classes. These findings demonstrate the potential of this approach for ecological monitoring. Ground-level imagery is easily obtained and accurate computational methods for habitat classification based on such data have many potential applications. To support use by practitioners, a simple web application is also provided that allows classification of uploaded images using the trained model.