利用二元逻辑回归算法，根据患者基本信息（年龄、BMI、有无手术史、有无既往史、是否吸烟、是否酗酒、有无PONV、有无晕动史）和镇静药物种类（B药、R药），对患者术中、术后 24h 的不良反应进行预判。具体步骤及matlab代码

步骤：

1. 收集患者基本信息和镇静药物种类以及对应的不良反应数据，构建数据集。

2. 对数据集进行预处理，包括缺失值处理、异常值处理、数据归一化等。

3. 将数据集分为训练集和测试集，通常采用70%的数据作为训练集，30%的数据作为测试集。

4. 使用二元逻辑回归算法进行训练，得到模型。

5. 对测试集进行预测，计算预测准确率和召回率等指标，评估模型性能。

6. 对新的患者进行预测，根据预测结果提前采取措施，减少不良反应的发生。

Matlab 代码：

% Load data
data = load('data.txt');
X = data(:, 1:8); % Features
y = data(:, 9); % Labels

% Preprocess data
X = normalize(X);

% Split data into training and testing sets
m = size(X, 1);
train_size = round(0.7 * m);
indices = randperm(m);
train_indices = indices(1:train_size);
test_indices = indices(train_size+1:end);
X_train = X(train_indices, :);
y_train = y(train_indices);
X_test = X(test_indices, :);
y_test = y(test_indices);

% Train logistic regression model
theta = logistic_regression(X_train, y_train);

% Predict test set labels
y_pred = predict(X_test, theta);

% Evaluate model performance
accuracy = sum(y_pred == y_test) / length(y_test);
precision = sum(y_pred & y_test) / sum(y_pred);
recall = sum(y_pred & y_test) / sum(y_test);
f1_score = 2 * precision * recall / (precision + recall);

% Display results
fprintf('Accuracy: %f\n', accuracy);
fprintf('Precision: %f\n', precision);
fprintf('Recall: %f\n', recall);
fprintf('F1 score: %f\n', f1_score);

% Predict new patient label
patient = [35, 23, 1, 0, 1, 0, 1, 0];
patient = (patient - mean(X)) ./ std(X);
label = predict(patient, theta);

% Logistic regression function
function theta = logistic_regression(X, y)
m = size(X, 1);
n = size(X, 2);
X = [ones(m, 1) X];
theta = zeros(n+1, 1);
alpha = 0.01;
num_iterations = 1000;
for i = 1:num_iterations
h = sigmoid(X * theta);
gradient = X' * (h - y) / m;
theta = theta - alpha * gradient;
end
end

% Sigmoid function
function g = sigmoid(z)
g = 1.0 ./ (1.0 + exp(-z));
end

% Predict function
function p = predict(X, theta)
m = size(X, 1);
X = [ones(m, 1) X];
p = sigmoid(X * theta) >= 0.5;
end