In addition to the interactive, education-focused meeting, YoungMD Connect (YMDC) Live offers support to qualifying applicants who submit an original research abstract to the YMDC Live Scholarship Program. These presentations often highlight early-career ophthalmic research focused on improving training access, triage efficiency, surgical decision-making, and longitudinal disease management. Across the five winning 2026 abstracts, the recurring theme was practical innovation: the use of low-cost visualization tools, validated clinical indicators, preoperative phenotyping, home monitoring, and predictive algorithms to make ophthalmic care more targeted and efficient.

Low-Cost Heads-Up Microsurgical Training

In his presentation, Bryan Adams, DO, identified a gap between advanced OR visualization systems and the limited tools available to trainees for at-home or wet-lab practice. Dr. Adams explained that traditional microscope setups and hobby microscopes create ergonomic limitations, such as forward leaning and potential neck/back strain. Therefore, he posed the question: Is there a way to develop anaglyph 3D red-cyan heads-up visualization for ophthalmologic microsurgical training?

Dr. Adams and collaborators created a simple heads-up display microscope using commercially available components. The system uses a small digital microscope that connects to an available screen such as that of a PC, laptop, mobile phone, or tablet. The device offers low-cost, 2D visualization but lacks depth perception.

Dr. Adams’ initial testing with the device demonstrated that the device is durable, versatile, and portable. Additionally, it can be used for practicing routine maneuvers in microsurgery. At the time of presentation, the research team was developing real-time 3D visualization with paired microscopes and red-cyan filtered glasses using anaglyph technology.

Dr. Adams concluded that while this system may be beneficial for practicing basic ophthalmic microsurgical techniques outside the OR, its major limitation was loss of brightness/color information, which matters in ophthalmology because surgery relies heavily on visual cues. This project positioned low-cost heads-up visualization as a potential educational tool for microsurgical practice while acknowledging that reduced image quality may limit its broader use.

Orbital Fracture Consultation Algorithms

Urjita Das, MD, presented research identifying clinical indicators and screening algorithms that can determine when orbital fractures require ophthalmic consultation and whether all consultations are clinically necessary.

Dr. Das and colleagues compared the sensitivity and specificity of four existing orbital fracture consultation screening algorithms: HOPE+CT, STOP, MEE, and UTH. The investigators examined data between 2015 and 2020 from a large orbital fracture consultation database. Of 11,986 ophthalmologic consultations identified during the study period, 814 patients with 913 eyes with orbital fractures were included in the analysis.

Their results demonstrated that 218 eyes (24%) required intervention; interventions excluded conservative measures such as rest, ice, head elevation, and sinus precautions. Topical antibiotic administration was the most common intervention (98 eyes; 10.7%), followed by periorbital laceration repair (47 eyes; 5.1%). Additionally, relative afferent pupillary defect, extraocular muscle entrapment, gaze restriction, periorbital laceration, and subconjunctival hemorrhage were significant predictors of intervention following ophthalmologic consultation.

The researchers found the STOP protocol to be the highest-sensitivity algorithm, resulting in a 29% reduction in hospital fracture consultations. From these results, Dr. Das concluded that most patients with orbital fractures did not require intervention following ophthalmologic consultation. Additionally, these results supported a case for prospectively using the STOP protocol for orbital fractures in the acute setting.

Orbital Compliance in Thyroid Eye Disease

Amy Shteyman, a fourth year medical student, presented research focused on using preoperative orbital compliance to predict outcomes after orbital fat decompression in stable-phase thyroid eye disease (TED). TED is known to cause orbital tissue proliferation, fat/muscle involvement, fibrosis, proptosis, and eyelid retraction.¹ Once symptoms had been stable for approximately 6 months, patients may be eligible for definitive surgical correction.

Shteyman and collaborators conducted a retrospective chart review of stable-phase patients with TED who underwent orbital fat decompression by a single surgeon. They measured orbital compliance preoperatively with a Hertel exophthalmometer. Ultimately, they hypothesized high-compliance orbits to reflect more pliable, fat-predominant disease; low-compliance orbits may have reflected more fibrosis. Forty-six orbits from 24 patients were included, comprising 15 orbits with low compliance and 31 with high orbital compliance. Results demonstrated that high compliance was associated with greater mean proptosis reduction than low compliance. Additionally, high-compliance patients showed greater variability in proptosis outcomes. Next steps will include an AI algorithm to evaluate the tissue differences between low- and high-compliance groups. This research determined that orbital compliance may help predict proptosis reduction after fat decompression, but high-compliance patients had less predictable outcomes.

Home Tonometry in Glaucoma Management

In her presentation, Yi-Hsien Yeh, MD, discussed clinical scenarios where home tonometry provided actionable data to guide medical and surgical interventions. Glaucoma is a longitudinal disease, but clinic visits provide only single IOP snapshots. Dr. Yeh explained that home tonometry may identify spikes, diurnal variation, and inadequate pressure control that can be missed during office visits.

This case series included chart reviews of patients who underwent home IOP monitoring with the iCare HOME2. Results demonstrated that patient A, who had ocular hypertension, exhibited normal OCT/visual field readings in clinic, but home IOP spikes measured above 30 mm Hg, leading to treatment with SLT. Patient B demonstrated progressive thinning of the retinal nerve fiber layer and worsening of the visual field despite consistent office IOPs of 9 to 12 mm Hg. This patient experienced IOP elevations in the evenings, prompting trabeculectomy with subsequent visual field stabilization. Patient C experienced persistent postoperative IOP swings with home monitoring after XEN Gel Stent implantation (AbbVie). Similarly, this patient underwent a trabeculectomy to improve IOP stability.

Overall, Dr. Yeh concluded that home tonometry provided actionable longitudinal data that could guide escalation from observation or medication to laser or surgery and could support postoperative monitoring.

AI Model for MIGS Selection

In the final scholarship presentation, Shawn Yuan, MD, identified a gap between controlled MIGS studies and real-world surgical decision-making. To address this, he introduced an early pilot study aimed at developing a preliminary AI-based predictive model capable of estimating preoperative patient characteristics and identifying optimal MIGS procedures to minimize the risk of glaucoma progression.

To begin, he analyzed a retrospective multicenter database of Canadian patients undergoing MIGS with 1 to 5 years of follow-up. Five surgeons from three academic sites were included, and each surgeon had access to a wide range of MIGS options. Clinical parameters included MIGS type, concurrent phacoemulsification, medical and surgical history, baseline visual acuity, IOP, and corneal thickness. Additionally, machine learning models were trained to identify whether patient characteristics predicted better outcomes with one MIGS device versus another.

Dr. Yuan completed preliminary analyses for 3,200 eyes undergoing trabecular meshwork bypass and gonioscopy-assisted transluminal trabeculotomy. The results demonstrated that among predictive approaches, logistic regression showed the most consistent performance metrics compared with more complex machine learning models. The AI model achieved moderate calibration at 1 year and meaningful predictive agreement beyond chance. To this end, Dr. Yuan concluded that this early pilot study demonstrated the potential for an AI-driven framework to support personalized MIGS selection and outcome prediction.

1. Khong JJ, McNab AA, Ebeling PR, et al. Pathogenesis of thyroid eye disease: review and update on molecular mechanisms. Br J Ophthalmol 2016;100:142–150.