Update app.py

app.py

@@ -1,4 +1,3 @@
-# app.py (7-feature aligned, server-safe)
 import gradio as gr
 import pandas as pd
 import joblib
@@ -14,8 +13,8 @@ warnings.filterwarnings("ignore")
 MODEL_PATH = "models/svm_pipeline.joblib"
 BG_PATH    = "data/bg.csv"
 
-# 模型最终需要的 7 个特征（顺序必须与训练一致）
-FEATURES = ["ALB", "TP", "TBA", "AST_ALT", "CREA", "PNI", "AAPR"]
+# 模型最终需要的 8 个特征（顺序必须与训练一致）
+FEATURES = ["ALB", "TP", "TBA", "AST_ALT", "CREA", "PNI", "AAPR", "DBIL"]
 
 # 加载模型与背景
 pipeline = joblib.load(MODEL_PATH)
@@ -52,15 +51,13 @@ def _render_force_plot(base_val: float, shap_1d: np.ndarray, feat_1d: np.ndarray
 
     # ——将特征值按需四舍五入，减少标签长度（可根据需要微调每列小数位）
     feat = np.asarray(feat_1d, dtype=float).copy()
-    # 这里给出一套通用规则：ALB/TP/CREA/PNI 一位小数；TBA 两位；AST_ALT 与 AAPR 两位
     round_map = {
-        "ALB": 2, "TP": 2, "TBA": 2, "AST_ALT": 2, "CREA": 1, "PNI": 1, "AAPR": 3
+        "ALB": 2, "TP": 2, "TBA": 2, "AST_ALT": 2, "CREA": 1, "PNI": 1, "AAPR": 3, "DBIL": 1
     }
     feat_rounded = [
         np.round(val, round_map.get(name, 2)) for val, name in zip(feat, fnames)
     ]
 
-    # 如需进一步减字，可把特征名缩写或去掉空格，这里保持原名
     shap.force_plot(
         base_val,
         np.asarray(shap_1d).reshape(-1),
@@ -80,7 +77,8 @@ def _render_force_plot(base_val: float, shap_1d: np.ndarray, feat_1d: np.ndarray
 def _coerce_float(x):
     return float(x) if x is not None and x != "" else np.nan
 
-def predict_and_explain(ALB, TP, TBA, AST_ALT, CREA, LYM, ALP, nsamples=200):
+
+def predict_and_explain(ALB, TP, TBA, AST_ALT, CREA, LYM, ALP, DBIL, nsamples=200):
     status = []
     try:
         # ---- 1) 取数并校验 ----
@@ -91,21 +89,24 @@ def predict_and_explain(ALB, TP, TBA, AST_ALT, CREA, LYM, ALP, nsamples=200):
         CREA    = _coerce_float(CREA)
         LYM     = _coerce_float(LYM)
         ALP     = _coerce_float(ALP)
+        DBIL    = _coerce_float(DBIL)
 
-        vals = [ALB, TP, TBA, AST_ALT, CREA, LYM, ALP]
+        vals = [ALB, TP, TBA, AST_ALT, CREA, LYM, ALP, DBIL]
         if any(np.isnan(v) for v in vals):
             return None, None, "Error: 所有输入必须为数值且不可缺失。"
 
         if ALP <= 0:
             return None, None, "Error: ALP 必须 > 0（用于计算 AAPR=ALB/ALP）。"
+        if DBIL < 0:
+            return None, None, "Error: DBIL 不能为负值。"
 
         # ---- 2) 衍生指标 ----
         PNI  = ALB + 5.0 * LYM
         AAPR = ALB / ALP
-        status.append(f"Derived: PNI={PNI:.1f}, AAPR={AAPR:.2f}")
+        status.append(f"Derived: PNI={PNI:.1f}, AAPR={AAPR:.3f}")
 
-        # ---- 3) 组装最终 7 特征并预测 ----
-        x_row = np.array([[ALB, TP, TBA, AST_ALT, CREA, PNI, AAPR]], dtype=np.float64)
+        # ---- 3) 组装最终 8 特征并预测 ----
+        x_row = np.array([[ALB, TP, TBA, AST_ALT, CREA, PNI, AAPR, DBIL]], dtype=np.float64)
 
         if hasattr(pipeline, "predict_proba"):
             classes_ = getattr(pipeline, "classes_", None)
@@ -162,7 +163,7 @@ def predict_and_explain(ALB, TP, TBA, AST_ALT, CREA, LYM, ALP, nsamples=200):
         except Exception as e_force:
             status.append(f"Force-plot failed: {repr(e_force)}; fallback=bar")
             order = np.argsort(np.abs(sv))[::-1]
-            topk = order[:min(7, sv.shape[0])]
+            topk = order[:min(len(FEATURES), sv.shape[0])]
             plt.close('all')
             fig = plt.figure(figsize=(8, 5), dpi=160)
             plt.barh(np.array(FEATURES)[topk], sv[topk])
@@ -176,19 +177,21 @@ def predict_and_explain(ALB, TP, TBA, AST_ALT, CREA, LYM, ALP, nsamples=200):
     except Exception as e:
         return None, None, f"Fatal error: {repr(e)}"
 
-# ====== 示例输入（仅 7 项 + nsamples）======
-example_values = [41.7, 64.9, 0.870, 0.890, 55.9, 1.95, 51.96, 200]
-
-
+# ====== 示例输入（8 项 + nsamples）======
+# 给定示例：
+# ALB=40.7, TP=61.7, TBA=4.9, AST/ALT=1.0, CREA=54.3, PNI=48.6, AAPR=0.473, DBIL=8.5
+# 由于前端输入仍为原始 6 项 + DBIL（PNI、AAPR 在后端计算），需“还原” LYM 与 ALP：
+# LYM=(PNI-ALB)/5=(48.6-40.7)/5=1.58； ALP=ALB/AAPR=40.7/0.473≈86.0465
+example_values = [40.7, 61.7, 4.9, 1.0, 54.3, 1.58, 86.0465, 8.5, 200]
 
 
 # ====== Gradio 界面 ======
 with gr.Blocks() as demo:
     gr.Markdown(
         "### Meige Risk Prediction (SVM) with SHAP Explanation\n"
-        "Please enter ALB, TP, TBA, AST/ALT, CREA, LYM, and ALP. The values for PNI and AAPR will be calculated automatically.\n\n"
+        "Please enter ALB, TP, TBA, AST/ALT, CREA, LYM, ALP, and DBIL.\n\n"
         "**Units**: ALB(g/L), TP(g/L), TBA(μmol/L), AST/ALT(ratio), CREA(μmol/L), "
-        "LYM(×10⁹/L), ALP(U/L)."
+        "LYM(×10⁹/L), ALP(U/L), DBIL(μmol/L)."
     )
 
     with gr.Row():
@@ -201,6 +204,7 @@ with gr.Blocks() as demo:
                 gr.Number(label="CREA (μmol/L)"),
                 gr.Number(label="LYM (×10⁹/L)"),
                 gr.Number(label="ALP (U/L)"),
+                gr.Number(label="DBIL (μmol/L)"),
             ]
             ns_slider = gr.Slider(100, 500, value=200, step=50, label="SHAP nsamples")
 
@@ -223,4 +227,4 @@ with gr.Blocks() as demo:
     )
 
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch()