MELD与MELD-Na评分Python实现准确性核查请求

问题背景 #

作为研究者，我正在用Python实现肝硬化评分的MELD和MELD-Na公式，代码可运行但部分结果与手动计算值不符，需要核查代码实现是否正确。

官方公式 #

MELD = 1.33（女性额外加） + [4.56 × loge(bilirubin)] + [0.82 × (137 – sodium)] – [0.24 × (137 – sodium) × loge(bilirubin)] + [9.09 × loge(INR)] + [11.14 × loge(creatinine)] + [1.85 × (3.5 – albumin)] – [1.83 × (3.5 – albumin) × loge(creatinine)] + 6

MELD-Na = MELD + 1.32 × (137-Na) – [0.033 × MELD × (137-Na)]

原实现代码 #

import pandas as pd
import math

def calculate_meld(bilirubin, sodium, INR, creatinine, albumin, Sex):
    if bilirubin <= 0 or sodium <= 0 or INR <= 0 or creatinine <= 0:
        return None  # Return None if any of the required parameters are non-positive
    
    if Sex.lower() == 'female':
        meld_score = (1.33 +
                      4.56 * math.log(bilirubin) + 
                      0.82 * (137 - sodium) - 
                      0.24 * (137 - sodium) * math.log(bilirubin) + 
                      9.09 * math.log(INR) + 
                      11.14 * math.log(creatinine) + 
                      1.85 * (3.5 - albumin) - 
                      1.83 * (3.5 - albumin) - 
                      1.83 * (3.5 - albumin) * math.log(creatinine) + 
                      6)
    else:
        meld_score = (4.56 * math.log(bilirubin) + 
                      0.82 * (137 - sodium) - 
                      0.24 * (137 - sodium) * math.log(bilirubin) + 
                      9.09 * math.log(INR) + 
                      11.14 * math.log(creatinine) + 
                      1.85 * (3.5 - albumin) - 
                      1.83 * (3.5 - albumin) - 
                      1.83 * (3.5 - albumin) * math.log(creatinine) + 
                      6)
    
    meld_score = round(meld_score)
    meld_score = max(6, min(meld_score, 40))  # Limiting the MELD score between 6 and 40
    return meld_score

def calculate_meld_na(meld, sodium):
    if meld is None:
        return None  # Return None if MELD score is None
    
    meld_na_score = meld + 1.32 * (137 - sodium) - 0.033 * meld * (137 - sodium)
    
    meld_na_score = round(meld_na_score)
    meld_na_score = max(6, min(meld_na_score, 40))  # Limiting the MELD-Na score between 6 and 40
    return meld_na_score

# Load data from Excel file
data = pd.read_excel('ICC NSQIP Data.xlsx')

# Drop rows with missing data
data.dropna(subset=['Total Bilirubin', 'Serum Sodium', 'INR', 'Serum Creatinine', 'Albumin', 'Sex'], inplace=True)

# Extract relevant columns
bilirubin = data['Total Bilirubin']
sodium = data['Serum Sodium']
INR = data['INR']
creatinine = data['Serum Creatinine']
albumin = data['Albumin']
Sex = data['Sex']

# Calculate MELD and MELD-Na scores
meld_scores = [calculate_meld(b, s, i, c, a, g) for b, s, i, c, a, g in zip(bilirubin, sodium, INR, creatinine, albumin, Sex)]
meld_scores = [score if score is not None else 0 for score in meld_scores]  # Replace None with 0

meld_na_scores = [calculate_meld_na(m, s) for m, s in zip(meld_scores, sodium)]

# Add MELD and MELD-Na scores to the DataFrame
data['MELD Score'] = meld_scores
data['MELD-Na Score'] = meld_na_scores

# Save the updated DataFrame to a new Excel file
data.to_excel('liver_data_updated_2.xlsx', index=False)

错误分析与修正 #

核心错误：公式项重复计算 #

原代码中calculate_meld函数的男女分支里，都多写了一行- 1.83 * (3.5 - albumin)，这在官方MELD公式中不存在。官方公式仅包含– [1.83 × (3.5 – albumin) × loge(creatinine)]这一项，多余的减项会导致计算结果偏低，与手动值不符。

修正后的calculate_meld函数 #

def calculate_meld(bilirubin, sodium, INR, creatinine, albumin, Sex):
    if bilirubin <= 0 or sodium <= 0 or INR <= 0 or creatinine <= 0:
        return None  # Return None if any of the required parameters are non-positive
    
    base_score = (4.56 * math.log(bilirubin) + 
                  0.82 * (137 - sodium) - 
                  0.24 * (137 - sodium) * math.log(bilirubin) + 
                  9.09 * math.log(INR) + 
                  11.14 * math.log(creatinine) + 
                  1.85 * (3.5 - albumin) - 
                  1.83 * (3.5 - albumin) * math.log(creatinine) + 
                  6)
    
    if Sex.lower() == 'female':
        meld_score = base_score + 1.33
    else:
        meld_score = base_score
    
    meld_score = round(meld_score)
    meld_score = max(6, min(meld_score, 40))  # Limiting the MELD score between 6 and 40
    return meld_score

其他验证点 #

• Python的math.log()默认是自然对数（loge），符合公式要求，无需修改。
• MELD-Na的公式实现正确，无逻辑错误。
• 数值截断（限制在6-40之间）符合临床评分规范。

额外优化建议 #

将None替换为0可能不符合临床逻辑，建议保留None或用NaN填充，避免后续统计误差：

meld_scores = [score if score is not None else pd.NA for score in meld_scores]

内容的提问来源于stack exchange，提问作者Not a Wrench