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
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
| """
ECG + 眼动双模态认知负荷分类器
基于论文发现:
- Mean NN (ECG) 是第3重要特征
- W分量 (ECG超低频) 是第4重要特征
- 瞳孔直径 是第5重要特征
去掉fNIRS后, 使用ECG+眼动实现3级分类:
- Underload (低负荷)
- Moderate (适中)
- Overload (过载)
"""
import numpy as np
from sklearn.ensemble import GradientBoostingClassifier
from sklearn.model_selection import cross_val_score, LeaveOneGroupOut
from sklearn.preprocessing import StandardScaler
from scipy.signal import find_peaks, butter, filtfilt, welch
from typing import Tuple
class CognitiveLoadClassifier:
"""ECG+眼动双模态认知负荷分类器"""
def __init__(self):
self.scaler = StandardScaler()
self.model = GradientBoostingClassifier(
n_estimators=150,
max_depth=4,
learning_rate=0.1,
subsample=0.8,
random_state=42
)
self.labels = ['低负荷', '适中', '过载']
def extract_ecg_features(self, ecg: np.ndarray, fs=250) -> np.ndarray:
"""提取ECG/HRV特征: Mean NN, RMSSD, LF/HF, W分量"""
peaks, _ = find_peaks(ecg, distance=150, prominence=0.5)
if len(peaks) < 10:
return np.zeros(4)
rr = np.diff(peaks) / fs
mean_nn = np.mean(rr)
rmssd = np.sqrt(np.mean(np.diff(rr)**2))
freqs, psd = welch(rr, fs=1/mean_nn, nperseg=min(len(rr), 64))
lf = np.sum(psd[(freqs >= 0.04) & (freqs < 0.15)])
hf = np.sum(psd[(freqs >= 0.15) & (freqs < 0.40)])
w = np.sum(psd[(freqs >= 0.01) & (freqs < 0.04)])
return np.array([mean_nn, rmssd, lf/(hf+1e-8), w])
def extract_eye_features(self, eye_data: np.ndarray, fs=30) -> np.ndarray:
"""提取眼动特征: 瞳孔直径, 眨眼率, 注视稳定性"""
eye_openness, gaze_x, gaze_y, pupil_d = eye_data.T
pupil_mean = np.mean(pupil_d)
pupil_var = np.std(pupil_d)
blink_threshold = 0.3
blink_events = np.diff((eye_openness < blink_threshold).astype(int))
blink_rate = np.sum(blink_events == -1) / (len(eye_openness) / fs / 60)
gaze_std = np.std(np.sqrt(gaze_x**2 + gaze_y**2))
return np.array([pupil_mean, pupil_var, blink_rate, gaze_std])
def extract_features(self, ecg, eye_data):
"""提取双模态特征"""
ecg_f = self.extract_ecg_features(ecg)
eye_f = self.extract_eye_features(eye_data)
return np.concatenate([ecg_f, eye_f])
def train(self, X, y, groups=None):
"""训练模型"""
X_scaled = self.scaler.fit_transform(X)
if groups is not None:
logo = LeaveOneGroupOut()
scores = cross_val_score(
self.model, X_scaled, y,
groups=groups, cv=logo, scoring='accuracy'
)
print(f"LOSO准确率: {scores.mean():.4f} ± {scores.std():.4f}")
self.model.fit(X_scaled, y)
return self.model
def predict(self, ecg, eye_data):
"""预测认知负荷等级"""
features = self.extract_features(ecg, eye_data).reshape(1, -1)
features_scaled = self.scaler.transform(features)
pred = self.model.predict(features_scaled)[0]
proba = self.model.predict_proba(features_scaled)[0]
return {
'level': int(pred),
'level_name': self.labels[int(pred)],
'confidence': float(max(proba)),
'probabilities': {self.labels[i]: float(p) for i, p in enumerate(proba)}
}
if __name__ == "__main__":
np.random.seed(42)
clf = CognitiveLoadClassifier()
n_per_class = 30
X_list, y_list = [], []
for level in range(3):
for _ in range(n_per_class):
mean_nn = 0.9 - level * 0.1 + np.random.normal(0, 0.05)
rmssd = 0.05 - level * 0.01 + np.random.normal(0, 0.01)
lf_hf = 1.0 + level * 0.5 + np.random.normal(0, 0.2)
w = 0.005 - level * 0.001 + np.random.normal(0, 0.002)
pupil = 3.5 + level * 0.3 + np.random.normal(0, 0.2)
pupil_var = 0.2 + level * 0.1 + np.random.normal(0, 0.05)
blink_rate = 15 - level * 3 + np.random.normal(0, 2)
gaze_std = 0.15 + level * 0.1 + np.random.normal(0, 0.03)
features = np.array([mean_nn, rmssd, lf_hf, w,
pupil, pupil_var, blink_rate, gaze_std])
X_list.append(features)
y_list.append(level)
X = np.array(X_list)
y = np.array(y_list)
groups = np.repeat(np.arange(3), n_per_class)
print("=== ECG+眼动双模态认知负荷分类器 ===\n")
clf.train(X, y, groups=np.repeat(np.arange(10), 9))
print("\n=== 预测示例 ===")
test_ecg = np.random.normal(0, 0.1, 15000)
test_ecg[np.random.choice(15000, 50, replace=False)] = 1.0
test_eye = np.column_stack([
np.random.normal(0.8, 0.05, 1800),
np.random.normal(0, 0.15, 1800),
np.random.normal(0, 0.15, 1800),
np.random.normal(3.8, 0.2, 1800)
])
result = clf.predict(test_ecg, test_eye)
print(f"预测: {result['level_name']} (置信度: {result['confidence']:.3f})")
print(f"概率: {result['probabilities']}")
|
