MultiModalPneumonia/Makefile at master · ELTE-DSED/MultiModalPneumonia · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
# Makefile — Multimodal Pneumonia Detection Pipeline
# Requires: .venv activated or use PYTHON variable below
PYTHON ?= python
SEED   := 42
export MPLCONFIGDIR := artifacts/.mpl_cache

# ─── Data pipeline ───────────────────────────────────────────────────────────
preprocess:
	bash scripts/run_data_pipeline.sh

preprocess_labs:
	bash scripts/run_lab_pipeline.sh

# ─── Pretraining ─────────────────────────────────────────────────────────────
pretrain:
	$(PYTHON) -m src.training.train_image_multilabel_pretrain

# ─── Image fine-tuning ───────────────────────────────────────────────────────
finetune_image:
	$(PYTHON) -m src.training.train_image_pneumonia_finetune --lr-head 5e-5 --lr-backbone 1e-5

# ─── Multimodal training (canonical run) ─────────────────────────────────────
finetune_multimodal:
	$(PYTHON) -m src.training.train_multimodal_pneumonia --lr-head 5e-5 --lr-backbone 1e-5

# ─── Clinical baselines ──────────────────────────────────────────────────────
train_clinical_lr:
	$(PYTHON) -m src.training.train_clinical_baseline

train_clinical_xgb:
	$(PYTHON) -m src.training.train_clinical_xgb

train_clinical: train_clinical_lr train_clinical_xgb

# ─── Evaluation ──────────────────────────────────────────────────────────────
bootstrap_delta:
	$(PYTHON) -m src.evaluation.bootstrap_eval \
	  --model-a artifacts/models/multimodal_pneumonia_densenet121_triage_u_ignore_temporal_stronger_lr_v3/test_predictions.csv \
	  --model-b artifacts/models/image_pneumonia_finetune_densenet121_u_ignore_temporal_stronger_lr_v3/test_predictions.csv \
	  --output-json artifacts/evaluation/bootstrap_multimodal_vs_image_stronger_lr_v3.json \
	  --n-bootstrap 2000 --seed $(SEED)

calibration:
	$(PYTHON) -m src.evaluation.calibration_analysis \
	  --output-dir artifacts/evaluation/calibration_stronger_lr_v3 \
	  --n-bins 10 --bootstrap --n-bootstrap 2000 \
	  --model "Image" artifacts/models/image_pneumonia_finetune_densenet121_u_ignore_temporal_stronger_lr_v3/test_predictions.csv \
	  --model "Multimodal" artifacts/models/multimodal_pneumonia_densenet121_triage_u_ignore_temporal_stronger_lr_v3/test_predictions.csv \
	  --model "Clinical Logistic" artifacts/models/clinical_baseline_u_ignore_temporal_strong_v2/test_predictions.csv \
	  --model "Clinical XGBoost" artifacts/models/clinical_xgb_u_ignore_temporal_strong_v2/test_predictions.csv
	$(PYTHON) -m src.evaluation.calibration_analysis \
	  --output-dir artifacts/evaluation/calibration_final \
	  --n-bins 10 --bootstrap --n-bootstrap 2000 \
	  --model "Image" artifacts/models/image_pneumonia_finetune_densenet121_u_ignore_temporal_stronger_lr_v3/test_predictions.csv \
	  --model "Multimodal" artifacts/models/multimodal_pneumonia_densenet121_triage_u_ignore_temporal_stronger_lr_v3/test_predictions.csv

dca:
	$(PYTHON) -m src.evaluation.decision_curve_analysis \
	  --output-dir artifacts/evaluation/dca \
	  --model "Image" artifacts/models/image_pneumonia_finetune_densenet121_u_ignore_temporal_stronger_lr_v3/test_predictions.csv \
	  --model "Multimodal" artifacts/models/multimodal_pneumonia_densenet121_triage_u_ignore_temporal_stronger_lr_v3/test_predictions.csv

feature_ablation:
	$(PYTHON) scripts/collect_feature_ablation_results.py

evaluate: bootstrap_delta calibration dca feature_ablation

# ─── SHAP ────────────────────────────────────────────────────────────────────
shap:
	$(PYTHON) scripts/generate_shap_clinical.py \
	  --model-dir artifacts/models/clinical_xgb_u_ignore_temporal_strong_v2 \
	  --feature-groups all

# ─── Publication report ───────────────────────────────────────────────────────
report:
	$(PYTHON) scripts/generate_publication_report.py

# ─── Testing ─────────────────────────────────────────────────────────────────
test:
	$(PYTHON) -m pytest tests/ -v --tb=short

# ─── Full pipeline ────────────────────────────────────────────────────────────
all: pretrain finetune_image finetune_multimodal train_clinical evaluate shap report

.PHONY: preprocess preprocess_labs pretrain finetune_image finetune_multimodal \
        train_clinical_lr train_clinical_xgb train_clinical bootstrap_delta \
        calibration dca feature_ablation evaluate shap report all test