shrishSVaidya/medgemma-1.5-mm-progression-module4

MedGemma 1.5 - Multiple Myeloma Progression Tracking (Module 4)

📈 Model Overview

This repository contains a PEFT/LoRA adapter fine-tuned on MedGemma 1.5 4B-IT. It is specifically designed to analyze raw clinical text, extract longitudinal M-Spike metrics (Serum Protein Electrophoresis), and assess if the data indicates rapid disease progression in Multiple Myeloma patients.

This adapter was developed as part of a broader agentic AI application catering to Multiple Myeloma patients. It acts as Module 4, operating alongside upstream risk assessment and vision modules to feed structured progression data into a RAG-enabled clinical dashboard.

🔗 Associated Code Repository

The complete source code for data preparation, training and validating this adapter, as well as the full Agentic-AI pipeline, can be found on GitHub: here

Base Model Dependency

This is an adapter model. It requires the base weights from Google's MedGemma 1.5 4B-IT.

⚠️ License and Terms of Use

• LoRA Adapter Weights: The adapter weights and associated code in this repository are open-sourced under the Apache 2.0 license.
• Base Model: To use this adapter, you must agree to the Google Health AI Developer Foundations Terms of Use to access the underlying MedGemma 1.5 weights.
• Clinical Disclaimer: This model is for educational and research purposes only. It is not a medical device, is not intended for clinical use, and should not be used to diagnose, treat, or offer medical advice for any disease or condition.

💻 How to Use

Because this model uses the MedGemma vision-language architecture, it is strictly recommended to load the model in 4-bit NF4 quantization and utilize a dummy image tensor to stabilize the cross-attention vision layers during text-only generation.

from transformers import AutoModelForImageTextToText, AutoProcessor, BitsAndBytesConfig
from peft import PeftModel
from PIL import Image
import torch

# 1. Load Base Model in 4-bit NF4
model_id = "google/medgemma-1.5-4b-it"
processor = AutoProcessor.from_pretrained(model_id)

quant_config = BitsAndBytesConfig(
    load_in_4bit=True,
    bnb_4bit_quant_type="nf4",
    bnb_4bit_use_double_quant=True,
    bnb_4bit_compute_dtype=torch.float16
)

base_model = AutoModelForImageTextToText.from_pretrained(
    model_id,
    device_map="auto",
    quantization_config=quant_config
)

# 2. Load this LoRA Adapter
model = PeftModel.from_pretrained(base_model, "shrish/medgemma-1.5-mm-progression-module4")

# 3. Format Prompt
patient_id = "" # fill in the patient id 
Timeline = "" # have the subsequent lab test metrics like: - Day -17: Platelets: 329.0 x10^9 cells/L, Hemoglobin: 6.7 mmol/L, Creatinine: 68.07 umol/L, M Protein: 2.98 g/dL, Calcium: 2.45 mmol/L
- Day 78: Calcium: 2.25 mmol/L, M Protein: 1.97 g/dL, Hemoglobin: 7.01 mmol/L, Platelets: 286.0 x10^9 cells/L, Creatinine: 53.92 umol/L, etc
prompt = f"Review the following longitudinal biomarker history for patient {patient_id}. Predict the disease trajectory: Is this patient showing biochemical progression toward Active Myeloma?
Timeline:{Timeline}"
messages = [{"role": "user", "content": prompt}]
formatted_prompt = processor.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)

inputs = processor(
    text=formatted_prompt, 
    return_tensors="pt", 
    padding=True
).to(model.device)
inputs.pop("token_type_ids", None)

# 4. Generate
with torch.no_grad():
    outputs = model.generate(**inputs, max_new_tokens=300, do_sample=False)

input_length = inputs["input_ids"].shape[1]
generated_tokens = outputs[0, input_length:]
pred_text = processor.decode(generated_tokens, skip_special_tokens=True).strip()

print("Model Prediction:\n", pred_text)