Overview

This case study illustrates how MediVerse's voice-driven VR interface streamlines exploration of pediatric leukemia datasets, eliminating the need for SQL or complex menu navigation. The dataset comprises two core tables: Patient Clinical Data and Patient Treatments.

Workflow

  1. Schema definition & data import. Define table schema and import CSVs.
  2. Voice-driven plotting. "Show a scatter plot of initial WBC versus initial blast counts for all patients by risk" generates a 3D plot coloured by risk stratification.
Scatter plot of initial WBC vs blast counts by risk stratification
Fig. 1. Initial WBC vs blast counts by risk stratification.
  1. Narrow to high-risk. "Show a scatter plot of initial WBC versus initial blast counts for high-risk patients" limits the view to the high-risk group.
  2. Colour by gender. "Show a scatter plot of initial WBC versus initial blast counts for high-risk patients coloured by gender" updates colouring for immediate visual comparison.
Gender-coloured scatter plot of high-risk patients
Fig. 2. Gender-coloured scatter plot of high-risk patients.
  1. Interactive table. "List all high-risk patients whose blast counts exceed 50 × 109" renders an in-VR table with patient ID, blast count, and clinical details.
Table of high-risk patients with blast counts exceeding threshold
Fig. 3. High-risk patients with blast counts exceeding 50 × 109.
  1. Clinical explanation. "Why spike in WBC counts for patient ALL101 post treatment initiation" triggers a synthesised clinical explanation highlighting possible biological mechanisms.
MediVerse clinical explanation for patient ALL101
Fig. 4. MediVerse clinical explanation for patient ALL101.

Results & impact

MediVerse enables seamless voice-driven transitions between scatter plots, interactive tables, and clinical narratives within a VR environment. Clinicians can explore gender differences in blast counts or compare treatment efficacy without leaving the immersive workspace.