Development and Evaluation of Aeyeconsult: A Novel Ophthalmology Chatbot Leveraging Verified Textbook Knowledge and GPT-4.

OBJECTIVE: There has been much excitement on the use of large language models (LLMs) such as ChatGPT in ophthalmology. However, LLMs are limited in that they are trained on unverified information and do not cite their sources. This paper highlights a new methodology to create a generative AI chatbot to answer eye care related questions which uses only verified ophthalmology textbooks as data and cites its sources. SETTING: Yale School of Medicine Department of Ophthalmology and Visual Science. DESIGN/METHODS: Aeyeconsult, an ophthalmology chatbot, was developed using GPT-4 (the LLM used to power the publicly available chatbot ChatGPT-4), LangChain, and Pinecone. Ophthalmology textbooks were processed into embeddings and stored in Pinecone. User queries were similarly converted, compared to stored embeddings, and GPT-4 generated responses. The interface was adapted from public code. Both Aeyeconsult and ChatGPT-4 were tested on the same 260 questions from OphthoQuestions.com, with the first response from Aeyeconsult and ChatGPT-4 recorded as the answer. RESULTS: Aeyeconsult outperformed ChatGPT-4 on the OKAP dataset, with 83.4% correct answers compared to 69.2% (p = 0.0118). Aeyeconsult also had fewer instances of no answer and multiple answers. Both systems performed best in General Medicine, with Aeyeconsult achieving 96.2% accuracy. Aeyeconsult's weakest performance was in Clinical Optics at 68.1%, but it still outperformed ChatGPT-4 in this category (45.5%). CONCLUSION: LLMs may be useful in answering ophthalmology questions but their trustworthiness and accuracy is limited due to training on unverified internet data and lack of source citation. We used a new methodology, using verified ophthalmology textbooks as source material and providing citations, to mitigate these issues, resulting in a chatbot more accurate than ChatGPT-4 in answering OKAPs style questions.

Photo-mediated ultrasound therapy for the treatment of retinal neovascularization in rabbit eyes.

OBJECTIVES: Retinal neovascularization (RNV) is the growth of abnormal microvessels on the retinal surface and into the vitreous, which can lead to severe vision loss. By combining relatively low-intensity ultrasound and nanosecond-pulse-duration laser, we developed a novel treatment method, namely photo-mediated ultrasound therapy (PUT), which holds a potential to remove RNV with minimal or no damage to the adjacent tissues. METHODS: RNV was created in both albino and pigmented rabbits (n = 10) through a single intravitreal injection with DL-α-aminoadipic acid. RNV was treated with PUT 8 weeks postinjection. After PUT treatment, animals were evaluated longitudinally for up to 6 weeks. Treatment outcomes were evaluated through fundus photography, red-free fundus photography, fluorescein angiography (FA), and histopathology. RESULTS: In both albino and pigmented rabbits, there were no leakage in the treatment area immediately after PUT treatment as demonstrated by FA, indicating the cessation of blood perfusion of the RNV in the treated area. The fluorescence leakage did not recover in albino rabbits during the 6-week posttreatment monitoring period, and only 9.9 ± 9.8% of the neovascularization remained at the end of 6 weeks. In the pigmented rabbits, the fluorescence leakage partially returned, but the level of leakage decreased over time during the 6-week posttreatment monitoring period, and only 10.8 ± 9.8% of the neovascularization remained at the end of 6 weeks. Histology demonstrated removal of vasculature without damage to the surrounding neurosensory retina. CONCLUSIONS: These results demonstrate that PUT could precisely remove RNV without damage to the surrounding neurosensory retina in both rabbit strains.

Impact of Contrast Echocardiography on Outcomes in Critically Ill Patients.

Limited data suggests ultrasound enhancing agent (UEA) use is associated with changes in clinical management and lower mortality in intensive care unit (ICU) patients. We conducted a retrospective observational study to determine if contrast echocardiography (vs non-contrast echocardiography) is associated with differences in length of stay (LOS) and subsequent resource utilization in the ICU setting. The Premier Healthcare Database (Charlotte, NC) was analyzed to identify patients receiving Definity vs. no use of contrast during the initial rest transthoracic echocardiogram (TTE) in an ICU setting. The primary outcomes of interest were subsequent TTE and transesophageal echocardiography (TEE) during the index hospitalization, and ICU LOS. Propensity scoring was used to statistically model treatment selection to minimize selection bias. A total of 1,538,864 patients from 773 hospitals were identified as undergoing resting TTE in the ICU with use of DEFINITY in 51,141 (3.3%) patients and no contrast agent use in 1,487,723 (96.7%) patients. After adjusting for patient, clinical, and hospital characteristics, patients in the Definity cohort were less likely to undergo a subsequent TTE or TEE as compared to those in the no contrast cohort (odds ratio = 0.704 for TTE, odds ratio = 0.841 for TEE; p < 0.0001 for both). Adjusted mean ICU LOS for the Definity cohort was shorter than that of the no contrast cohort (4.59 vs 4.15 days, p < 0.0001). In conclusion, Definity-enhanced echocardiography in the ICU setting (in comparison with non-contrast TTE) is associated with lower rates of subsequent TTE and TEE during the index hospitalization, and shorter ICU LOS.

Long-term multimodal imaging characterization of persistent retinal neovascularization using DL-alpha-aminoadipic acid in pigmented and white rabbits.

PURPOSE: Intravitreal (IVT) injection of DL-alpha-aminoadipic acid (AAA) is a new animal model for retinal neovascularization (RNV) reported in rabbits. This study performs longitudinal multimodal imaging for up to 1 year to evaluate DL-AAA RNV in both New Zealand white (NZW) rabbits and Dutch-Belted pigmented (DBP) rabbits. METHOD: Detailed characterization and quantification of this model were performed in these two strains in 32 eyes by optical coherence tomography (OCT), fundus photography, and fluorescein angiography (FA) for up to 16 weeks following DL-AAA administration in 32 eyes and up to 52 weeks in 5 eyes. H & E histology was also performed in these two strains 8 weeks after injection of DL-AAA. RESULT: RNV was successfully generated using 50 µL 80 mM DL-AAA solution for DBP rabbits and 80 µL 80 mM DL-AAA for NZW rabbits. The incidence of persistent vascular leakage is 100% (15/15) for DBP rabbits and 70.6% (12/17) for NZW rabbits at 16 weeks. Complications with NZW rabbits ultimately decreased the efficiency in NZW rabbits to 58.8% (10/17) of NZW rabbits getting persistent (to 16 weeks) vascular leakage without ocular complications as compared with 100% (15/15) in DBP rabbits. Five eyes (2 DBP and 3 NZW) were selected from those demonstrating RNV at 16 weeks and were monitored for up to 52 weeks. All 5 demonstrated persistent RNV to 52 weeks. Quantification of the mean leakage area (MLA) in DBP rabbits is more accurate than in NZW rabbits since the reduced contrast between the leakage and background in NZW rabbits makes it more challenging to quantify. CONCLUSION: DL-AAA can induce persistent and quantifiable RNV in both DBP and NZW rabbits. DBP rabbits have a higher success rate, lower required volume of DL-AAA, and more accurate method for quantification that could be more desirable.

Photo-Mediated Ultrasound Therapy for the Treatment of Corneal Neovascularization in Rabbit Eyes.

Purpose: Corneal neovascularization (CNV) is the invasion of new blood vessels into the avascular cornea, leading to reduced corneal transparency and visual acuity, impaired vision, and even blindness. Current treatment options for CNV are limited. We developed a novel treatment method, termed photo-mediated ultrasound therapy (PUT), that combines laser and ultrasound, and we tested its feasibility for treating CNV in a rabbit model. Methods: A suture-induced CNV model was established in New Zealand White rabbits, which were randomly divided into two groups: PUT and control. For the PUT group, the applied light fluence at the corneal surface was estimated to be 27 mJ/cm2 at 1064-nm wavelength with a pulse duration of 5 ns, and the ultrasound pressure applied on the cornea was 0.43 MPa at 0.5 MHz. The control group received no treatment. Red-free photography and fluorescein angiography were utilized to evaluate the efficiency of PUT. Safety was evaluated by histology and immunohistochemistry. For comparison with the PUT safety results, conventional laser photocoagulation (LP) treatment was performed with standard clinical parameters: 532-nm continuous-wave (CW) laser with 0.1-second pulse duration, 450-mW power, and 75-µm spot size. Results: In the PUT group, only 1.8% ± 0.8% of the CNV remained 30 days after treatment. In contrast, 71.4% ± 7.2% of the CNV remained in the control group after 30 days. Safety evaluations showed that PUT did not cause any damage to the surrounding tissue. Conclusions: These results demonstrate that PUT is capable of removing CNV safely and effectively in this rabbit model. Translational Relevance: PUT can remove CNV safely and effectively.

Photo-mediated Ultrasound Therapy to Treat Retinal Neovascularization.

This report describes a novel therapeutic technique called photo-mediated ultrasound therapy (PUT). PUT applies synchronized short pulse duration (nanosecond) laser and ultrasound burst on targeted tissue, offering high-precision localized treatment. PUT is based on controlled induction and promotion of micro-cavitation activity in the target tissue. PUT is able to safely and effectively treat retinal neovascularization in rabbits with persistent nonperfusion up to 4 weeks after PUT in the choroidal vasculature.Clinical Relevance- PUT can selectively remove retinal angiogenesis in clinically-relevant disease models in humansized eyes (rabbit) without damaging surrounding tissue.

Extracting and classifying diagnosis dates from clinical notes: A case study.

Myeloproliferative neoplasms (MPNs) are chronic hematologic malignancies that may progress over long disease courses. The original date of diagnosis is an important piece of information for patient care and research, but is not consistently documented. We describe an attempt to build a pipeline for extracting dates with natural language processing (NLP) tools and techniques and classifying them as relevant diagnoses or not. Inaccurate and incomplete date extraction and interpretation impacted the performance of the overall pipeline. Existing lightweight Python packages tended to have low specificity for identifying and interpreting partial and relative dates in clinical text. A rules-based regular expression (regex) approach achieved recall of 83.0% on dates manually annotated as diagnosis dates, and 77.4% on all annotated dates. With only 3.8% of annotated dates representing initial MPN diagnoses, additional methods of targeting candidate date instances may alleviate noise and class imbalance.

The Effect of Laser and Ultrasound Synchronization in Photo-Mediated Ultrasound Therapy.

OBJECTIVE: Photo-mediated ultrasound therapy (PUT) is a novel, non-invasive, agent-free, highly selective, and precise anti-vascular technique. PUT removes microvessels through promoting cavitation activity precisely in targeted microvessels by applying synchronized nanosecond laser pulses and ultrasound bursts. The synchronization between laser and ultrasound is critical to the outcome of PUT. METHODS: Through theoretical simulation and experimental study, the effect of synchronization between laser pulses and ultrasound bursts on cavitation activity during PUT is evaluated. RESULTS: By using a theoretical model, we found that cavitation activity was enhanced when laser pulses and ultrasound bursts were synchronized such that the produced photoacoustic wave overlaid the rarefactional phase of the ultrasound wave. This finding was then verified through in vitro studies where cavitation was monitored by using a passive cavitation detector. Furthermore, we demonstrated that the in vivo treatment outcome of PUT in rabbits was directly related to the synchronization between laser and ultrasound. The anti-vascular effect could only be observed when laser and ultrasound were properly synchronized in vivo. CONCLUSION: PUT is more efficient when the laser-induced photoacoustic wave overlays the rarefactional phase of the ultrasonic wave. SIGNIFICANCE: This is a systematic study to investigate the synchronization effect of PUT, which would be significant for further understanding the mechanism and further improving the treatment efficiency of PUT.

Long-Term Outcomes and Associations With Major Adverse Limb Events After Peripheral Artery Revascularization.

BACKGROUND: Long-term cardiovascular and limb outcomes after revascularization for peripheral artery disease and, in particular, prognosis after post-procedure major adverse limb events (MALE) are not well-studied. OBJECTIVES: This study sought to describe outcomes after peripheral revascularization and assess relationships between post-procedure MALE hospitalization and subsequent events. METHODS: Patients undergoing peripheral artery revascularization between January 1, 2009, and September 30, 2015, in the Premier Healthcare Database were examined for the co-primary outcomes of interest, composite myocardial infarction (MI) or stroke and composite major amputation or peripheral revascularization. Multivariable adjusted Cox proportional hazards models with post-procedure MALE hospitalization included as a time-dependent covariate were developed to estimate hazard ratios for outcomes. RESULTS: Among 393,017 revascularized patients followed for a median of 2.7 years (interquartile range: 1.3 to 4.4 years), the cumulative incidence of MI or stroke was 9.8% and that of major amputation or peripheral revascularization was 41.9%. A total of 50,750 patients (12.9%) had at least 1 post-procedure MALE hospitalization. In time-dependent covariate adjusted models, post-procedure MALE hospitalization was associated with greater risk of subsequent MI or stroke (hazard ratio: 1.34; 95% confidence interval: 1.28 to 1.40) and major amputation or peripheral revascularization (hazard ratio: 8.13; 95% confidence interval: 7.96 to 8.29). After peripheral revascularization with or without post-procedure MALE hospitalization, risk of limb events increased rapidly post-procedure and more slowly after the first year, whereas cardiac risk increased steadily during follow-up. CONCLUSIONS: Revascularized peripheral artery disease patients face earlier limb and later cardiovascular ischemic risk that is heightened among patients with post-procedure MALE hospitalization. Increased provider awareness of these long-term risks may guide efforts to improve post-procedural outcomes.