sighh

app/services/context_generator.py

@@ -235,19 +235,27 @@ class ContextGenerator:
         return metrics
 
     def _detect_thresholds(self) -> Tuple[Optional[Dict], Optional[Dict]]:
-        """Detect VT1 and VT2 thresholds"""
+        """Detect VT1 and VT2 thresholds (matching notebook logic)"""
+        # VT1: First index where carb burn > fat burn AND remains higher
         condition = self.pnoe_df["CHO_smoothed"] > self.pnoe_df["FAT_smoothed"]
         crossover_indices = condition[condition].index
 
         vt1 = None
         if len(crossover_indices) > 0:
-            vt1_idx = crossover_indices[0]
-            vt1_row = self.pnoe_df.loc[vt1_idx]
-            vt1 = {
-                "HeartRate": vt1_row["HR(bpm)_smoothed"],
-                "Speed": vt1_row["Speed"],
-                "Time": vt1_row["T(sec)"],
-            }
+            # Find first crossover where carbs remain higher for the rest
+            for idx in crossover_indices:
+                if all(
+                    self.pnoe_df.loc[idx:]["CHO_smoothed"]
+                    > self.pnoe_df.loc[idx:]["FAT_smoothed"]
+                ):
+                    vt1_idx = idx
+                    vt1_row = self.pnoe_df.loc[vt1_idx]
+                    vt1 = {
+                        "HeartRate": vt1_row["HR(bpm)_smoothed"],
+                        "Speed": vt1_row["Speed"],
+                        "Time": vt1_row["T(sec)"],
+                    }
+                    break
 
         ve_slope = self.pnoe_df["VE(l/min)_smoothed"].diff()
         second_derivative = ve_slope.diff()
@@ -745,9 +753,13 @@ class ContextGenerator:
         """Calculate detailed metrics for each heart rate zone based on actual data"""
         import math
 
-        # Get zone boundaries
-        fat_max_idx = self.pnoe_df["FAT_smoothed"].idxmax()
-        optimal_row = self.pnoe_df.loc[fat_max_idx]
+        # Get zone boundaries - use optimal fat burning zone (highest fat:carb ratio)
+        # matching notebook logic
+        self.pnoe_df["fat_carb_ratio"] = self.pnoe_df["FAT_smoothed"] / (
+            self.pnoe_df["CHO_smoothed"] + 0.00000001
+        )
+        optimal_fat_idx = self.pnoe_df["fat_carb_ratio"].idxmax()
+        optimal_row = self.pnoe_df.loc[optimal_fat_idx]
 
         # Detect VT1 and VT2
         vt1 = pnoe_metrics.get("vt1")