The future is bright: artificial intelligence for trainee medical officers in Australia and New Zealand.

Given their frontline role in Australia and Aotearoa New Zealand (ANZ) healthcare, trainee medical officers (TMOs) will play a crucial role in the development and use of artificial intelligence (AI) for clinical care, ongoing medical education and research. As 'digital natives', particularly those with technical expertise in AI, TMOs should also be leaders in informing the safe uptake and governance of AI within ANZ healthcare as they have a practical understanding of its associated risks and benefits. However, this is only possible if a culture of broad collaboration is instilled while the use of AI in ANZ is still in its initial phase.

Transorbital Endoscopic Approach to the Foramen Rotundum for Infraorbital Nerve Stripping.

PURPOSE: To develop and evaluate a transorbital endoscopic approach to the foramen rotundum to excise the maxillary nerve and infraorbital nerve branch. METHODS: Cadaveric dissection study of 10 cadaver heads (20 orbits). This technique is predicated upon 1) an inferior orbital fissure release to facilitate access to the orbital apex and 2) the removal of the posterior maxillary wall to enter the pterygopalatine fossa (PPF). Angulations along the infraorbital nerve were quantified as follows: the first angulation was measured between the orbitomaxillary segment within the orbital floor and the pterygopalatine segment suspended within the PPF, while the second angulation was taken between the pterygopalatine segment and maxillary nerve as it exited the foramen rotundum. With refinement of the technique, the minimum amount of posterior maxillary wall removal was quantified in the final 5 cadaver heads (10 orbits). RESULTS: The mean distance from the inferior orbital rim to the foramen rotundum was 45.55 ± 3.24 mm. The first angulation of the infraorbital nerve was 133.10 ± 16.28 degrees, and the second angulation was 124.95 ± 18.01 degrees. The minimum posterior maxillary wall removal to reach the PPF was 11.10 ± 2.56 mm (vertical) and 11.10 ± 2.08 mm (horizontal). CONCLUSIONS: The transorbital endoscopic approach to an en bloc resection of the infraorbital nerve branch up to its maxillary nerve origin provides a pathway to the PPF. This is relevant for nerve stripping in the context of perineural spread. Other applications include access to the superior portion of the PPF in selective biopsy cases or in concurrent orbital pathology.

Imaging Features of Invasive Fungal Rhinosinusitis: A Systematic Review.

Fungal rhinosinusitis (FRS) includes non-invasive and invasive subtypes with the latter having significant morbidity and mortality. This systematic review aims to identify the imaging features most correlated with invasive fungal rhinosinusitis (IFRS) and present a checklist of these features to aid diagnosis. PubMed, Embase, CENTRAL, and Science Direct were searched from inception to May 2023, in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines. Primary research articles published in English describing the imaging features of IFRS were included. The systematic review was conducted in accordance with the PRISMA guidelines. Forty-eight articles were identified for inclusion. Six studies examined radiological features in acute invasive fungal rhinosinusitis (AIFRS), and 9 studies of chronic invasive fungal rhinosinusitis (CIFRS). A majority of studies did not specify whether IFRS cases were acute or chronic. On CT, bony erosion and mucosal thickening were the most common features. Other features include nasal soft tissue thickening, nasal cavity opacification, opacification of the affected sinus, and perisinus soft tissue infiltration. Extra-sinus extension was commonly observed on MRI, most often invading intraorbitally and intracranially. Other sites of extra-sinus extension included the cavernous sinus, pterygopalatine fossa, infratemporal fossa, masticator space, and facial soft tissue. IFRS is a condition with potential for high morbidity and mortality. Several radiological features are highly suggestive of IFRS. Early identification of high-risk radiological features using a checklist may aid prompt diagnosis and early treatment. Future research investigating the radiological differentiation between IFRS and other significant pathology including bacterial orbital cellulitis would be beneficial.

A case study in applying artificial intelligence-based named entity recognition to develop an automated ophthalmic disease registry.

PURPOSE: Advances in artificial intelligence (AI)-based named entity extraction (NER) have improved the ability to extract diagnostic entities from unstructured, narrative, free-text data in electronic health records. However, there is a lack of ready-to-use tools and workflows to encourage the use among clinicians who often lack experience and training in AI. We sought to demonstrate a case study for developing an automated registry of ophthalmic diseases accompanied by a ready-to-use low-code tool for clinicians. METHODS: We extracted deidentified electronic clinical records from a single centre's adult outpatient ophthalmology clinic from November 2019 to May 2022. We used a low-code annotation software tool (Prodigy) to annotate diagnoses and train a bespoke spaCy NER model to extract diagnoses and create an ophthalmic disease registry. RESULTS: A total of 123,194 diagnostic entities were extracted from 33,455 clinical records. After decapitalisation and removal of non-alphanumeric characters, there were 5070 distinct extracted diagnostic entities. The NER model achieved a precision of 0.8157, recall of 0.8099, and F score of 0.8128. CONCLUSION: We presented a case study using low-code artificial intelligence-based NLP tools to produce an automated ophthalmic disease registry. The workflow created a NER model with a moderate overall ability to extract diagnoses from free-text electronic clinical records. We have produced a ready-to-use tool for clinicians to implement this low-code workflow in their institutions and encourage the uptake of artificial intelligence methods for case finding in electronic health records.

Morganella morganii and Enterococcus faecalis endophthalmitis following intravitreal injection.

BACKGROUND: Endophthalmitis following intravitreal injection is a potentially devastating complication of anti-VEGF injections. Post-injection endophthalmitis due to Enterococcus faecalis is rare, and no previous case of Morganella morganii endophthalmitis after intravitreal injection has been reported. CASE PRESENTATION: We present the first reported case of Morganella morganii and Enterococcus faecalis endophthalmitis after intravitreal injection in an immunocompetent patient in the absence of recent ocular surgery. Our patient presented with hand movement visual acuity one day after anti-VEGF injection and demonstrated no clinical improvement despite repeated intravitreal ceftazidime and vancomycin injections. A decision was made to proceed with early vitrectomy given failure of intravitreal antibiotics. Visual acuity improved to 6/90 at 12 weeks after vitrectomy without any evidence of disease recurrence. CONCLUSIONS: Post-injection endophthalmitis due to concurrent Morganella morganii and Enterococcus faecalis infections can have visually devastating consequences despite repeated empirical and targeted intravitreal antibiotics. Lack of clinical improvement following intravitreal antibiotics should warrant consideration of early vitrectomy. Our experience is a pertinent reminder of the ever-growing threat of uncommon and multi-resistant bacteria that must be considered when treating infections such as post-injection endophthalmitis.

Optimising vitrectomy operation note coding with machine learning.

BACKGROUND: The accurate encoding of operation notes is essential for activity-based funding and workforce planning. The aim of this project was to evaluate the procedural coding accuracy of vitrectomy and to develop machine learning, natural language processing (NLP) models that may assist with this task. METHODS: This retrospective cohort study involved vitrectomy operation notes between a 21-month period at the Royal Adelaide Hospital. Coding of procedures were based on the Medicare Benefits Schedule (MBS)-the Australian equivalent to the Current Procedural Terminology (CPT®) codes used in the United States. Manual encoding was conducted for all procedures and reviewed by two vitreoretinal consultants. XGBoost, random forest and logistic regression models were developed for classification experiments. A cost-based analysis was subsequently conducted. RESULTS: There were a total of 1724 procedures with individual codes performed within 617 vitrectomy operation notes totalling $1 528 086.60 after manual review. A total of 1147 (66.5%) codes were missed in the original coding that amounted to $736 539.20 (48.2%). Our XGBoost model had the highest classification accuracy (94.6%) in the multi-label classification for the five most common procedures. The XGBoost model was the most successful model in identifying operation notes with two or more missing codes with an AUC of 0.87 (95% CI 0.80-0.92). CONCLUSIONS: Machine learning has been successful in the classification of vitrectomy operation note encoding. We recommend a combined human and machine learning approach to clinical coding as automation may facilitate more accurate reimbursement and enable surgeons to prioritise higher quality clinical care.

Evaluating the Ability of Open-Source Artificial Intelligence to Predict Accepting-Journal Impact Factor and Eigenfactor Score Using Academic Article Abstracts: Cross-sectional Machine Learning Analysis.

BACKGROUND: Strategies to improve the selection of appropriate target journals may reduce delays in disseminating research results. Machine learning is increasingly used in content-based recommender algorithms to guide journal submissions for academic articles. OBJECTIVE: We sought to evaluate the performance of open-source artificial intelligence to predict the impact factor or Eigenfactor score tertile using academic article abstracts. METHODS: PubMed-indexed articles published between 2016 and 2021 were identified with the Medical Subject Headings (MeSH) terms "ophthalmology," "radiology," and "neurology." Journals, titles, abstracts, author lists, and MeSH terms were collected. Journal impact factor and Eigenfactor scores were sourced from the 2020 Clarivate Journal Citation Report. The journals included in the study were allocated percentile ranks based on impact factor and Eigenfactor scores, compared with other journals that released publications in the same year. All abstracts were preprocessed, which included the removal of the abstract structure, and combined with titles, authors, and MeSH terms as a single input. The input data underwent preprocessing with the inbuilt ktrain Bidirectional Encoder Representations from Transformers (BERT) preprocessing library before analysis with BERT. Before use for logistic regression and XGBoost models, the input data underwent punctuation removal, negation detection, stemming, and conversion into a term frequency-inverse document frequency array. Following this preprocessing, data were randomly split into training and testing data sets with a 3:1 train:test ratio. Models were developed to predict whether a given article would be published in a first, second, or third tertile journal (0-33rd centile, 34th-66th centile, or 67th-100th centile), as ranked either by impact factor or Eigenfactor score. BERT, XGBoost, and logistic regression models were developed on the training data set before evaluation on the hold-out test data set. The primary outcome was overall classification accuracy for the best-performing model in the prediction of accepting journal impact factor tertile. RESULTS: There were 10,813 articles from 382 unique journals. The median impact factor and Eigenfactor score were 2.117 (IQR 1.102-2.622) and 0.00247 (IQR 0.00105-0.03), respectively. The BERT model achieved the highest impact factor tertile classification accuracy of 75.0%, followed by an accuracy of 71.6% for XGBoost and 65.4% for logistic regression. Similarly, BERT achieved the highest Eigenfactor score tertile classification accuracy of 73.6%, followed by an accuracy of 71.8% for XGBoost and 65.3% for logistic regression. CONCLUSIONS: Open-source artificial intelligence can predict the impact factor and Eigenfactor score of accepting peer-reviewed journals. Further studies are required to examine the effect on publication success and the time-to-publication of such recommender systems.

Does Twitter improve the dissemination of ophthalmology scientific publications?

PURPOSE: To determine whether Twitter improves dissemination of ophthalmology scientific publications METHODS: Data were collected on articles published on PubMed between the years 2016 and 2021 (inclusive) and identified with the word "ophthalmology". Twitter performance metrics, including the number of tweets, number of likes, and number of retweets were collected from Twitter using the publicly available scientific API. Machine learning and descriptive statistics were used to outline Twitter performance metrics. RESULTS: The number of included articles was 433710. The percentage of articles that were in the top quartile for citation count, which had ≥1 tweet was 34.4% (number 437/1270). Conversely, the percentage of articles that were in the top quartile for citation count, which had 0 tweets was 27.8% (number 12023/43244). When machine learning was used to predict Twitter performance metrics an AUROC of 0.78 was returned. This was associated with an accuracy of 0.97 CONCLUSION: This study has shown preliminary evidence to support that Twitter may improve the dissemination of scientific ophthalmology publications.

Overcoming the impact of physiologic tremors in ophthalmology.

PURPOSE: Ophthalmic surgery involves the manipulation of micron-level sized structures such as the internal limiting membrane where tactile sensation is practically absent. All humans have physiologic tremors that are of low amplitude and not discernible to the naked eye; they do not adversely affect the majority of the population's daily functioning. However, during microsurgery, such tremors can be problematic. In this review, we focus on the impact of physiological tremors on ophthalmic microsurgery and offer a comparative discussion on the impact of such tremors on other surgical specialties. METHODS: A single investigator used the MEDLINE database (via PubMed) to search for and identify articles for inclusion in this systematic review. Ten key factors were identified as potentially having an impact on tremor amplitude: beta-blockers, muscle fatigue, robotic systems, handheld tools/micromanipulators, armrests/wrist supports, caffeine, diet, sleep deprivation, consuming alcohol, and workouts (exercise). These key terms were then searched using the advanced Boolean search tool and operators (i.e., AND, OR) available on PubMed: (*keyword*) AND (surgeon tremor OR microsurgery tremor OR hand steadiness OR simulator score). RESULTS: Ten studies attempted to quantify the baseline severity of operator physiologic tremor. Approximately 89% of studies accessing the impact of tremors on performance in regards to surgical metrics reported an improvement in performance compared to 57% of studies concluding that tremor elimination was of benefit when considering procedural outcomes. CONCLUSIONS: Robotic technology, new instruments, exoskeletons, technique modifications, and lifestyle factors have all demonstrated the potential to assist in overcoming tremors in ophthalmology.

Impact of cardiometabolic factors on retinal vasculature: A 3 × 3, 6 × 6 and 8 × 8-mm ocular coherence tomography angiography study.

BACKGROUND: Ocular coherence tomography angiography (OCTA) is available in varying size and resolution. We sought to characterise associations of cardiometabolic factors with retinal microvascular changes using 3 × 3, 6 × 6 and 8 × 8-mm OCTA scans to determine differences in detection with varying scan size. METHODS: Cross-sectional study of 247 cardiovascular patients from a single-centre tertiary-care hospital. Demographic, comorbidity and medication data were obtained. Patients underwent 3 × 3, 6 × 6 and 8 × 8-mm macula OCTA scanning using Carl Zeiss CIRRUS HD-OCT Model 5000. Angioplex and AngioTool software was used to quantify vascular parameters in the superficial capillary plexus. RESULTS: Increasing age, hypertension, dyslipidaemia, diabetes, chronic kidney disease, coronary artery disease and peripheral vascular disease were associated with reductions in vessel density, vessel perfusion, average vessel length and/or junction density in 3 × 3-mm OCTA (P < .05 for all). Conversely, smoking was associated with increased vessel density, vessel length and junction density in 3 × 3-mm OCTA (P < .05 for all). Associations of vessel abnormalities with cardiometabolic factors were progressively weakened and statistically attenuated in 6 × 6 and 8 × 8-mm OCTA scans. In multivariate analyses, dyslipidaemia remained an independent predictor of reduced vessel density, average vessel length and junction density (P < .05). CONCLUSIONS: Cardiometabolic factors are associated with multiple retinal microvascular changes in 3 × 3-mm OCTA scans. These associations were weakened and progressively attenuated in OCTA scans of larger 6 × 6 and 8 × 8-mm size. These findings advance our understanding of microcirculatory dysfunction and may have future implications for the screening and management of patients with cardiometabolic risk factors. Additional studies are required to further investigate these important associations.

Triaging ophthalmology outpatient referrals with machine learning: A pilot study.

IMPORTANCE: Triaging of outpatient referrals to ophthalmology services is required for the maintenance of patient care and appropriate resource allocation. Machine learning (ML), in particular natural language processing, may be able to assist with the triaging process. BACKGROUND: To determine whether ML can accurately predict triage category based on ophthalmology outpatient referrals. DESIGN: Retrospective cohort study. PARTICIPANTS: The data of 208 participants was included in the project. METHODS: The synopses of consecutive ophthalmology outpatient referrals at a tertiary hospital were extracted along with their triage categorizations. Following pre-processing, ML models were applied to determine how accurately they could predict the likely triage categorization allocated. Data was split into training and testing sets (75%/25% split). ML models were tested on an unseen test set, after development on the training dataset. MAIN OUTCOME MEASURE: Area under the receiver operator curve (AUC) for category one vs non-category one classification. RESULTS: For the main outcome measure, convolutional neural network (CNN) provided the best AUC (0.83) and accuracy on the test set (0.81), with the artificial neural network (AUC 0.81 and accuracy 0.77) being the next best performing model. When the CNN was applied to the classification task of identifying which referrals should be allocated a category one vs category two vs category three priority, a lower accuracy was achieved (0.65). CONCLUSIONS AND RELEVANCE: ML may be able to accurately assist with the triaging of ophthalmology referrals. Future studies with data from multiple centres and larger sample sizes may be beneficial.

The Observational Physician and surGEon Automobile Response (TOP GEAR) survey.

OBJECTIVE: To determine whether surgeons and junior doctors intending to pursue careers in surgery are more likely to purchase more expensive vehicles and to replace them sooner than colleagues of similar seniority pursuing non-surgical careers. DESIGN AND SETTING: Survey of practising medical officers at an Australian tertiary referral hospital. MAIN OUTCOME MEASURES: Car value; proportion of doctors who bought their car new; median time to replacement of vehicle. RESULTS: 154 doctors participated in the survey (17% response rate). 49% were interns, residents or unaccredited registrars, 18% were accredited registrars or fellows, and 31% were consultants; 40% of respondents were surgical trainees or consultants. 59% of surgical trainees and consultants purchased their car new, compared with 38% of non-surgical doctors (P = 0.013); 52% of doctors in the junior surgeon group purchased their car new, compared with 28% of non-surgeon junior doctors (P = 0.019). Median car value was $16 500 (IQR, $9350-37 000) for surgeons and $8500 (IQR, $4400-14 100) for non-surgeons (P < 0.001); 30% of surgeons owned cars valued at more than $50 000, compared with 6% of non-surgeons (P = 0.025). The median time to replacement was 5-7 years for surgeons and 7-10 years for non-surgeons (P < 0.001). CONCLUSIONS: Surgeons more frequently purchase their cars new and replace their cars earlier than non-surgeons, and the median value of their vehicles is higher. These findings were consistent across all levels of seniority.