Evaluating the translation of implementation science to clinical artificial intelligence: a bibliometric study of qualitative research.

Introduction: Whilst a theoretical basis for implementation research is seen as advantageous, there is little clarity over if and how the application of theories, models or frameworks (TMF) impact implementation outcomes. Clinical artificial intelligence (AI) continues to receive multi-stakeholder interest and investment, yet a significant implementation gap remains. This bibliometric study aims to measure and characterize TMF application in qualitative clinical AI research to identify opportunities to improve research practice and its impact on clinical AI implementation. Methods: Qualitative research of stakeholder perspectives on clinical AI published between January 2014 and October 2022 was systematically identified. Eligible studies were characterized by their publication type, clinical and geographical context, type of clinical AI studied, data collection method, participants and application of any TMF. Each TMF applied by eligible studies, its justification and mode of application was characterized. Results: Of 202 eligible studies, 70 (34.7%) applied a TMF. There was an 8-fold increase in the number of publications between 2014 and 2022 but no significant increase in the proportion applying TMFs. Of the 50 TMFs applied, 40 (80%) were only applied once, with the Technology Acceptance Model applied most frequently (n = 9). Seven TMFs were novel contributions embedded within an eligible study. A minority of studies justified TMF application (n = 51,58.6%) and it was uncommon to discuss an alternative TMF or the limitations of the one selected (n = 11,12.6%). The most common way in which a TMF was applied in eligible studies was data analysis (n = 44,50.6%). Implementation guidelines or tools were explicitly referenced by 2 reports (1.0%). Conclusion: TMFs have not been commonly applied in qualitative research of clinical AI. When TMFs have been applied there has been (i) little consensus on TMF selection (ii) limited description of selection rationale and (iii) lack of clarity over how TMFs inform research. We consider this to represent an opportunity to improve implementation science's translation to clinical AI research and clinical AI into practice by promoting the rigor and frequency of TMF application. We recommend that the finite resources of the implementation science community are diverted toward increasing accessibility and engagement with theory informed practices. The considered application of theories, models and frameworks (TMF) are thought to contribute to the impact of implementation science on the translation of innovations into real-world care. The frequency and nature of TMF use are yet to be described within digital health innovations, including the prominent field of clinical AI. A well-known implementation gap, coined as the "AI chasm" continues to limit the impact of clinical AI on real-world care. From this bibliometric study of the frequency and quality of TMF use within qualitative clinical AI research, we found that TMFs are usually not applied, their selection is highly varied between studies and there is not often a convincing rationale for their selection. Promoting the rigor and frequency of TMF use appears to present an opportunity to improve the translation of clinical AI into practice.

A genome-wide relay of signalling-responsive enhancers drives hematopoietic specification.

Developmental control of gene expression critically depends on distal cis-regulatory elements including enhancers which interact with promoters to activate gene expression. To date no global experiments have been conducted that identify their cell type and cell stage-specific activity within one developmental pathway and in a chromatin context. Here, we describe a high-throughput method that identifies thousands of differentially active cis-elements able to stimulate a minimal promoter at five stages of hematopoietic progenitor development from embryonic stem (ES) cells, which can be adapted to any ES cell derived cell type. We show that blood cell-specific gene expression is controlled by the concerted action of thousands of differentiation stage-specific sets of cis-elements which respond to cytokine signals terminating at signalling responsive transcription factors. Our work provides an important resource for studies of hematopoietic specification and highlights the mechanisms of how and where extrinsic signals program a cell type-specific chromatin landscape driving hematopoietic differentiation.

Prediction of visual function from automatically quantified optical coherence tomography biomarkers in patients with geographic atrophy using machine learning.

Geographic atrophy (GA) is a vision-threatening manifestation of age-related macular degeneration (AMD), one of the leading causes of blindness globally. Objective, rapid, reliable, and scalable quantification of GA from optical coherence tomography (OCT) retinal scans is necessary for disease monitoring, prognostic research, and clinical endpoints for therapy development. Such automatically quantified biomarkers on OCT are likely to further elucidate structure-function correlation in GA and thus the pathophysiological mechanisms of disease development and progression. In this work, we aimed to predict visual function with machine-learning applied to automatically acquired quantitative imaging biomarkers in GA. A post-hoc analysis of data from a clinical trial and routine clinical care was conducted. A deep-learning automated segmentation model was applied on OCT scans from 476 eyes (325 patients) with GA. A separate machine learning prediction model (Random Forest) used the resultant quantitative OCT (qOCT) biomarkers to predict cross-sectional visual acuity under standard (VA) and low luminance (LLVA). The primary outcome was regression coefficient (r2) and mean absolute error (MAE) for cross-sectional VA and LLVA in Early Treatment Diabetic Retinopathy Study (ETDRS) letters. OCT parameters were predictive of VA (r2 0.40 MAE 11.7 ETDRS letters) and LLVA (r2 0.25 MAE 12.1). Normalised random forest feature importance, as a measure of the predictive value of the three constituent features of GA; retinal pigment epithelium (RPE)-loss, photoreceptor degeneration (PDR), hypertransmission and their locations, was reported both on voxel-level heatmaps and ETDRS-grid subfields. The foveal region (46.5%) and RPE-loss (31.1%) had greatest predictive importance for VA. For LLVA, however, non-foveal regions (74.5%) and PDR (38.9%) were most important. In conclusion, automated qOCT biomarkers demonstrate predictive significance for VA and LLVA in GA. LLVA is itself predictive of GA progression, implying that the predictive qOCT biomarkers provided by our model are also prognostic.

Objective quantification of vitreous haze on optical coherence tomography scans: no evidence for relationship between uveitis and inflammation in multiple sclerosis.

BACKGROUND AND PURPOSE: The occurrence of intermediate uveitis, which is characterized by the presence of vitreous haze (VH), in patients with multiple sclerosis (MS) may be a sign of coexistent inflammatory central nervous system (CNS) disease activity. Using an automated algorithm to quantify VH on optical coherence tomography (OCT) scans, the aim was to investigate whether VH in MS patients is associated with signs of inflammatory CNS disease activity. METHODS: Vitreous haze was quantified on OCT macular volume scans of 290 MS patients and 85 healthy controls (HCs). The relationship between VH and clinical, retinal OCT and magnetic resonance imaging parameters of inflammatory disease activity was investigated using generalized estimating equations. RESULTS: Mean VH scores did not differ between patients and HCs (P = 0.629). Six patients (2.1%) showed values higher than the highest of the controls by HCs. VH scores did not differ between the different disease types or between eyes with and without a history of optic neuritis (P = 0.132). VH was not associated with inner nuclear layer volume on OCT (P = 0.233), cerebral T2 lesion load on magnetic resonance imaging (P = 0.416) or the development of new relapses (P = 0.205). CONCLUSION: In this study, OCT-based automated VH estimation did not detect increased vitreous inflammation in MS patients compared to HCs and did not find an association with CNS inflammatory burden.

Trichloroethylene and its metabolite TaClo lead to degeneration of substantia nigra dopaminergic neurones: Effects in wild type and human A30P mutant α-synuclein mice.

Parkinson's disease (PD) is characterised pathologically by degeneration of the dopaminergic (DA) neurones of the substantia nigra pars compacta (SNpc) and the presence of α-synuclein containing Lewy body inclusions. Trichloroethylene (TCE) has been suggested as a potential environmental chemical that may contribute to the development of PD, via conversion to the neurotoxin, 1-Trichloromethyl-1,2,3,4-tetrahydro-ß-carboline (TaClo). We investigated the effect of an 8 week exposure to TCE or TaClo on wild type and, as an experimental model of PD, A30P mutant α-synuclein overexpressing mice using a combination of behaviour and pathology. TCE or TaClo exposure caused significant DA neuronal loss within the SNpc in both wild type and transgenic mice. Cell numbers were lower in A30P animals than wild type, however, no additive effect of TCE or TaClo exposure and A30P overexpression was found. TCE or TaClo did not appear to lead to acceleration of motor or cognitive deficits in either wild type or A30P mutant mice, potentially because of the modest reductions of DA neuronal number in the SNpc. Our results do however suggest that TCE exposure could be a possible factor in development of PD like changes following exposure.

Optimizing OCT acquisition parameters for assessments of vitreous haze for application in uveitis.

Detection and evaluation of inflammatory activity in uveitis is essential to the management of the condition, and yet continues to be largely dependent on subjective clinical measures. Optical coherence tomography (OCT) measurement of vitreous activity is an alternative to clinical vitreous haze scoring and has passed a number of early validation studies. In this study we aimed to evaluate the impact of 'operator factors' on the variability of the technique as part of the validation process, and to help evaluate its suitability for 'real world' use. Vitreous haze index was calculated as a ratio between the reflectivity of the vitreous and of the outer retina in each scan. Different scanning conditions were tested and their effect on the measurement is reported. Our results show that the 'quantitative imaging' technique of OCT-measured vitreous activity had good reliability in normal subjects under a range of 'real world' conditions, such as when the operator changes the averaging value. The technique was however vulnerable to highly inaccurate focussing or abnormal downward displacement of the image. OCT-based quantification of vitreous activity is a promising alternative to current subjective clinical estimates, with sufficient 'tolerance' to be used in routine clinical practice as well as clinical trials.

An overview of the clinical applications of optical coherence tomography angiography.

Optical coherence tomography angiography (OCTA) has emerged as a novel, non-invasive imaging modality that allows the detailed study of flow within the vascular structures of the eye. Compared to conventional dye angiography, OCTA can produce more detailed, higher resolution images of the vasculature without the added risk of dye injection. In our review, we discuss the advantages and disadvantages of this new technology in comparison to conventional dye angiography. We provide an overview of the current OCTA technology available, compare the various commercial OCTA machines technical specifications and discuss some future software improvements. An approach to the interpretation of OCTA images by correlating images to other multimodal imaging with attention to identifying potential artefacts will be outlined and may be useful to ophthalmologists, particularly those who are currently still unfamiliar with this new technology. This review is based on a search of peer-reviewed published papers relevant to OCTA according to our current knowledge, up to January 2017, available on the PubMed database. Currently, many of the published studies have focused on OCTA imaging of the retina, in particular, the use of OCTA in the diagnosis and management of common retinal diseases such as age-related macular degeneration and retinal vascular diseases. In addition, we describe clinical applications for OCTA imaging in inflammatory diseases, optic nerve diseases and anterior segment diseases. This review is based on both the current literature and the clinical experience of our individual authors, with an emphasis on the clinical applications of this imaging technology.

The use of transdermal optical coherence tomography to image the superficial temporal arteries.

PurposeThe diagnosis of Giant Cell Arteritis (GCA) is an area of major challenge. This is the first reported use of the directed use of transdermal optical coherence tomography (OCT) to image the superficial temporal artery (STA).MethodsThis proof of concept study used a commercially available transdermal OCT instrument to identify and image the STA in eight patients (suspected GCA, confirmed GCA, and in healthy controls). Three cases are presented to demonstrate the preliminary imaging findings.ResultsIn all eight cases the STA was identified. Imaging findings from three cases are presented. A hyper-reflective signal was seen, which distinguishes the artery from vein. In two cases, a ratio of band thickness (BT) to arterial lumen diameter (ALD) could be calculated (BT : ALD ratio) where the whole circumference of the artery was imaged.DiscussionUsing dermal OCT to image the temporal arteries is a novel concept. With ongoing advances in resolution, penetration, and blood flow detection; this non-invasive technology warrants further investigation to determine its role in Giant Cell Arteritis.

Impact of optical coherence tomography scanning density on quantitative analyses in neovascular age-related macular degeneration.

PurposeTo assess the influence of varying B-scan frame-sampling densities on retinal thickness and volume measurements from spectral domain optical coherence tomography (OCT) in eyes with neovascular age-related macular degeneration (AMD).MethodsVolume OCT data (512 × 128 macular cube over 6 × 6 mm) were collected from 39 eyes with neovascular AMD. All 128 B-scans in each image set were manually segmented, allowing quantification of the neurosensory retina, subretinal fluid (SRF), subretinal hyperreflective material (SRHM), and pigment epithelium detachment (PED). Thickness maps were generated for less dense subsets of scans, ranging from every other (64 B-scans) to every 64th (2 B-scans). For each less dense subset, foveal central subfield thickness and total macular volume (TMV) were compared with values obtained using all 128 scans (considered the reference).ResultsFor each parameter, the mean absolute difference compared with the reference increased with reducing B-scan density. However, these differences did not reach statistical significance until frame-sampling density was reduced to every eighth scan (ie, 16 B-scans spaced 375 µm apart) for neurosensory retina, and every fourth scan (ie, 32 B-scans spaced 188 µm apart) for SRF, SRHM, and PED. For neurosensory retina, the mean (% error) and maximum (% error) absolute differences in TMV were 0.02 mm3 (0.24%) and 0.06 mm3 (0.79%), respectively. Similarly, at a density of 32 B-scans, mean and maximum differences for SRF were 0.004 mm3 (3.47%) and 0.02 mm3 (22.22%), respectively. The mean differences for SRHM and PED were 0.01 mm3 (8.03%) and 0.01 mm3 (4.04%), respectively.ConclusionsA minimum of 16 equally spaced B-scans, covering a 6 × 6 mm area, appears necessary to generate retinal thickness measurements similar to those obtained using all 128 B-scans in eyes with choroidal neovascularization (CNV). When considering other CNV lesion features, a minimum of 16 B-scans for SRF and PED, and 32 B-scans for SRHM are required to generate volume maps similar to ground-truth values. These findings may have implications for the design of acquisition and grading protocols for clinical trials using OCT in neovascular AMD.

Fingolimod: therapeutic mechanisms and ocular adverse effects.

Fingolimod is an oral immunomodulating drug used in the management of relapsing-remitting multiple sclerosis (RRMS). We aim to review the published literature on ocular manifestations of fingolimod therapy and their possible underlying mechanisms. The therapeutic effects of fingolimod are mediated via sphingosine receptors, which are found ubiquitously in various organs, including lymphoid cells, central nervous system, cardiac myocytes, and smooth muscle cells. Fingolimod-associated macular oedema (FAME) is the most common ocular side effect but retinal haemorrhages and retinal vein occlusion can occur. The visual consequences appear to be mild and, in cases of FAME, resolution is often attained with discontinuation of therapy. However, in cases of retinal vein occlusion, discontinuation of fingolimod alone may not be sufficient and intra-vitreal therapy may be required. We also propose a pragmatic service pathway for monitoring patients on fingolimod therapy, which includes stratifying them by risk and visual acuity.

The UK Neovascular AMD Database Report 3: inter-centre variation in visual acuity outcomes and establishing real-world measures of care.

PurposeInternational variations in visual acuity (VA) outcomes of eyes treated for neovascular age-related macular degeneration (nAMD) are well-documented, but intra-country inter-centre regional variations are not known. These data are important for national quality outcome indicators. We aimed to determine intra-country and inter-centre regional variations in outcomes for treatment of nAMD.Patients and methodsProspective multicentre national database study of 13 UK centres that treated patients according to a set protocol (three loading doses, followed by Pro-Re-Nata retreatment). A total of 5811 treatment naive eyes of 5205 patients received a total of 36 206 ranibizumab injections over 12 months.ResultsMean starting VA between centres varied from 48.9 to 59.9 ETDRS letters. Mean inter-centre VA change from baseline to 12 months varied from +6.9 letters to -0.6 letters (mean of +2.5 letters). The proportion of eyes achieving VA of 70 letters or more varied between 21.9 and 48.7% at 12 months. Median number of injections (visits) at each centre varied from 5 to 8 (9 to 12), with an overall median of 6 (11). Age, starting VA, number of injections, and visits, but not gender were significantly associated with variation in these VA outcomes (P<0.01). Significant variation between centres persisted even after adjusting for these factors.ConclusionThere are modest differences in VA outcomes between centres in the UK. These differences are influenced, but not completely explained, by factors such as patient age, starting VA, number of injections, and visits. These data provide an indication of the VA outcomes that are achievable in real-world settings.

Structural changes of the choroid in sarcoid- and tuberculosis-related granulomatous uveitis.

AIM: The aim of this study is to characterise the choroidal features of patients diagnosed with sarcoid- and tuberculosis (TB)-associated granulomatous uveitis using spectral domain optical coherence tomography (OCT). METHODS: Twenty-seven patients (27 eyes) diagnosed with sarcoid- (13 eyes) and TB (14 eyes)-related uveitis were included in this retrospective, cross-sectional study. Over a six-month period, patients diagnosed with sarcoid and TB granulomatous uveitis were scanned using enhanced depth imaging OCT. Clinical and demographical characteristics were recorded, including the method of diagnosis, disease activity, site of inflammation (anterior or posterior), treatments, and visual acuity (VA). Manual segmentation of the choroidal layers was performed using custom image analysis software. RESULTS: The main outcome measure was OCT-derived thickness measurements of the choroid and choroidal sublayers (Haller's large vessel and Sattler's medium vessel layers) at the macula region. The ratio of Haller's large vessel to Sattler's medium vessel layer was significantly different at the total macula circle in eyes diagnosed with TB uveitis (1.47 (=140.71/95.72 µm)) compared with sarcoid uveitis (1.07 (=137.70/128.69 µm)) (P=0.001). A thinner choroid was observed in eyes with a VA ≥0.3 LogMAR (Snellen 6/12; 198.1 µm (interquartile range (IQR)=147.0-253.4 µm) compared with those with VA <0.3 LogMAR (292.4 µm (IQR=240.1-347.6 µm)) at the total macula circle (P=0.004). At the foveal central subfield, the median choroidal thickness was 336.8 µm (IQR=272.3-375.4 µm) in active compared with 239.3 µm (IQR=195.3-330.9 µm) in quiescent disease (P=0.04). CONCLUSION: A disproportionately enlarged Sattler's layer may indicate a diagnosis of sarcoid-related uveitis, and choroidal thickening may be a feature of active granulomatous uveitis.

Mitochondrial dysfunction in Parkinson's disease.

Parkinson's disease (PD) is a progressive, neurodegenerative condition that has increasingly been linked with mitochondrial dysfunction and inhibition of the electron transport chain. This inhibition leads to the generation of reactive oxygen species and depletion of cellular energy levels, which can consequently cause cellular damage and death mediated by oxidative stress and excitotoxicity. A number of genes that have been shown to have links with inherited forms of PD encode mitochondrial proteins or proteins implicated in mitochondrial dysfunction, supporting the central involvement of mitochondria in PD. This involvement is corroborated by reports that environmental toxins that inhibit the mitochondrial respiratory chain have been shown to be associated with PD. This paper aims to illustrate the considerable body of evidence linking mitochondrial dysfunction with neuronal cell death in the substantia nigra pars compacta (SNpc) of PD patients and to highlight the important need for further research in this area.

Daily, monthly and seasonal variation in PSA levels and the association with weather parameters.

PSA levels have shown daily and seasonal variation, although data are conflicting regarding the season with higher PSA levels and the clinical relevance of this. We assessed the correlation of total PSA levels with meteorological data on a daily, weekly, monthly and seasonal basis. Data from 53,224 men aged 45-74 years, with an initial PSA <10.0 ng ml(-1) were correlated with temperature (°C), duration of bright sunshine (hours) and rainfall (mm). There was seasonal variation in PSA levels, with median PSA being higher in spring compared with other seasons (1.18 vs 1.10 ng ml(-1), P = 0.004). Seasonal variation was not apparent when PSA levels were age-adjusted (P = 0.112). Total PSA was not correlated with daily, weekly or monthly hours of sunshine, rainfall or mean temperature. In contrast, age-adjusted PSA varied with weekday, with higher PSA levels on Thursday and Friday compared with other days (1.16 vs 1.10 ng ml(-1), respectively). On multivariate analysis, only age predicted for PSA levels >3.0 ng ml(-1). In conclusion, PSA levels did show seasonal variation, although there was no direct correlation between PSA and any meteorological parameter. The degree of seasonal variation is small and the decision to proceed to prostate biopsy should be independent of season or weather parameters.

Reassessing Vermisporium (Amphisphaeriaceae), a genus of foliar pathogens of eucalypts.

The genus Vermisporium presently accommodates 13 species, 11 of which are associated with leaf spots of eucalypts in the Southern Hemisphere. Vermisporium is chiefly distinguished from Seimatosporium (Amphisphaeriaceae) on the basis of a short exogenous basal appendage, and the absence of a recognisable apical appendage. Due to the increasing importance of these species in native forests, and confusion pertaining to their taxonomy, a revision of the genus was undertaken based on fresh collections and dried herbarium specimens. Results from DNA sequence data analyses of the nrDNA-ITS and 28S nrRNA genes for species of Vermisporium indicated the genus to be a synonym of Seimatosporium. New combinations are introduced in Seimatosporium for several species: S. acutum, S. biseptatum, S. brevicentrum, S. obtusum, S. orbiculare, S. verrucisporum and S. walkeri. An updated key to all species occurring on eucalypts is also provided.

Imaging chorioretinal vascular disease.

Since its first description more than 40 years ago, fluorescein angiography had a crucial role in the diagnosis and management of chorioretinal vascular disorders such as neovascular age-related macular degeneration. Although fluorescein angiography permits visualization of the retinal microcirculation in exquisite detail, visualization of the choroidal circulation is more limited. Moreover, fluorescein angiography provides only minimal information regarding the functional consequences of vascular disease and allows, at best, only semi-quantitative assessment of retinal thickness. In recent years, the development of other chorioretinal imaging modalities, such as indocyanine green angiography, fundus autofluorescence, and optical coherence tomography (OCT), has addressed many of these issues. In particular, OCT has become an integral tool for vitreoretinal specialists as it allows high-resolution cross-sectional images of the neurosensory retina to be obtained in a non-invasive manner. The latest generation of commercial OCT technology-spectral domain OCT-offers high-speed scanning that allows complete coverage of the macular area, generation of three-dimensional retinal reconstructions, and precise image registration for inter-visit comparisons. The high speed of spectral domain OCT also facilitates B-scan averaging, which reduces speckle noise artefact and allows unparalleled visualization of the outer retina and choroid. In the near future, further advances in OCT technology (eg Doppler OCT) are likely to dramatically enhance the diagnosis and management of patients with chorioretinal vascular disease.

Accuracy of retinal thickness measurements obtained with Cirrus optical coherence tomography.

AIM: To report the frequency and severity of retinal thickness measurement errors in a Fourier domain optical coherence tomography (FDOCT) device, Cirrus OCT. METHODS: Data from 209 eyes undergoing Cirrus OCT imaging with the Macular Cube protocol were collected. For each eye, the position of the automated retinal boundary lines used by the Cirrus OCT software for thickness calculations was assessed using a 6-point categorical scale. The presence of errors was correlated with various parameters including: retinal morphological features and disease diagnosis. RESULTS: Errors of retinal boundary detection were observed in 57.5% of eyes but were severe in only 9.6% of eyes. The identification of subretinal fluid, subretinal tissue, pigment epithelium detachment or a diagnosis of choroidal neovascularisation was associated with more severe errors. Retinal cysts or a diagnosis of retinal vascular disease were less likely to be associated with significant error. CONCLUSIONS: Retinal thickness measurement errors appear to occur less frequently with Fourier domain OCT (Cirrus OCT), but segmentation errors remain a concern, particularly in assessment of eyes with structurally complex retinal disease. With the recent release of multiple FDOCT systems, assessment of segmentation error may be an important factor in determining the relative merits of these systems.