We describe DS@GT's submission to the eRisk 2026 Task 1 challenge on conversational depression screening. The system interviews LLM personas simulating individuals with varying depression profiles, producing a Beck Depression Inventory II (BDI-II) score and four key symptoms without directly asking sensitive mental health questions.
Our pipeline evolved through three stages: an initial monolithic single-model prototype, a baseline multi-agent architecture separating conversational interviewing from BDI-II scoring under a coordinating orchestration layer, and a final hybrid configuration replacing the paid GPT-5-nano interviewer with the open-source Gemma 27B.
To offset the model's weaker reasoning and instruction-following, the hybrid adds three algorithmic components: a precomputed dialogue tree standardizing interview openers and follow-ups, a reliability-weighted consensus aggregation inspired by the Weaver framework, and a cluster-based imputation step for unprobed symptoms.
We submitted three fully automated runs across all 20 personas, with Run 1 from the paid baseline and Runs 2 and 3 from the hybrid. Hybrid Run 3 achieved an ADODL of 0.9063, ranking 3rd among all complete-submission runs and placing DS@GT 2nd among the 21 teams overall, outperforming our paid baseline Run 1 (0.8841) at roughly one-quarter of the per-persona API cost.
These results support our central hypothesis that with sufficient algorithmic supervision, a weaker open-source model can compete with a stronger proprietary model in the conversational interviewer role. Our source code is available at GitHub.
Blogger's Review: This research highlights the potential of open-source models in specific applications, particularly in mental health screening. With effective algorithmic supervision, open-source technologies can achieve efficient conversational interactions even with limited resources, warranting further exploration of their applicability in other domains.