3. Data, Science & AI

Analyzes agent-user interaction transcripts to identify context network maintenance needs and guidance improvements for AI systems.

Verifies claims in AI-generated content against external sources to detect and correct hallucinations, ensuring factual accuracy.

Processes ebooks to extract concepts and entities with citation traceability, classify by type/layer, and synthesize across collections.

Calculates KDIGO Acute Kidney Injury staging for ICU patients using MIMIC-IV data and clinical criteria for nephrology research and renal monitoring.

Identifies Sepsis-3 patients (SOFA >= 2 + suspected infection) in MIMIC-IV for cohort studies and outcome research.

Calculates LODS for ICU patients using MIMIC-IV data to assess organ dysfunction across six systems with weighted scoring.

Calculates SAPS-II scores for ICU patients using MIMIC-IV data to assess severity, predict mortality, and enable benchmarking.

Calculates APACHE III scores from MIMIC-IV ICU patient data for mortality prediction, severity assessment, and risk-adjusted outcome analysis.

Calculates OASIS score for ICU patients using MIMIC-IV data to predict mortality with minimal lab requirements.

Navigates MIMIC-IV table relationships, join patterns, and identifier hierarchy for accurate data linkage and temporal joins.

Calculates norepinephrine-equivalent doses from vasopressor data in MIMIC-IV for hemodynamic support quantification and shock severity assessment.

Maps equivalent concepts between MIMIC-IV and eICU medical databases to enable cross-database research and validation.

Identifies suspected infection events in MIMIC-IV by analyzing antibiotic administration and culture timing, supporting Sepsis-3 criteria and infection research.

Calculates SOFA scores from MIMIC-IV data to assess organ dysfunction, sepsis severity, and mortality risk in ICU patients.

Extracts and calculates Glasgow Coma Scale (GCS) from MIMIC-IV ICU patient data for neurological assessment and trauma severity scoring.

Identifies first ICU stays and hospital admissions in MIMIC-IV data to support cohort selection, exclude readmissions, and build patient cohorts for medical research.

Enables programmatic querying of clinical datasets (MIMIC-IV, eICU) via Python API for data analysis and research.

Estimates baseline serum creatinine from MIMIC-IV data for AKI staging and renal function assessment in clinical research.

Guides researchers in avoiding critical data analysis pitfalls in clinical research using MIMIC-IV and eICU databases, addressing biases like immortal time and selection bias.

Calculates Systemic Inflammatory Response Syndrome (SIRS) criteria from MIMIC-IV ICU patient data for sepsis research and historical comparison.

Enables AI agents to evaluate, log, and observe LLM performance using Braintrust with SQL-based log queries and evaluation scripts.

Manages creation, modification, and analysis of AI agent structures including orchestrators, skills, and sub-agents.

Creates multi-host podcast episodes with AI-generated cloned voices, custom intro music, and audio transitions.

Provides AI-powered image recognition and kid-friendly narration via text-to-speech for educational content.