MySleepStage.Core 1.1.0

MySleepStage.Core

睡眠分期分析组件，提供数据清洗、数据预处理、数据平滑和睡眠分期分类功能。

功能特性

• 数据清洗: 剔除无效数据，包括无人状态超过阈值或移动幅度过大后的数据
• 数据预处理: 差值计算、加窗平滑、完整平滑（移动平均、高斯滤波、样条插值）
• 数据平滑: 独立的平滑服务，支持移动平均、高斯滤波、样条插值、自适应阈值过滤
• 睡眠分期: 支持离床、清醒、浅睡、深睡四种睡眠状态识别

安装

dotnet add package MySleepStage.Core

使用方法

1. 注册服务

using MySleepStage.Core.Extensions;

// 使用默认配置
services.AddSleepStageAnalysis();

// 或自定义配置
services.AddSleepStageAnalysis(options =>
{
    options.WindowSizeMinutes = 5;
    options.MonitoringStartHour = 22;
    options.MonitoringEndHour = 8;
    options.DefaultPreprocessingMode = PreprocessingMode.Auto;
    
    // 平滑配置
    options.MovingAverageWindowSize = 12;
    options.GaussianSigma = 2.5;
    options.SplineSmoothingParameter = 150;
    options.AdaptiveSensitivityFactor = 0.7;
});

2. 数据清洗

using MySleepStage.Core.Services;
using MySleepStage.Core.Models;

public class MyService
{
    private readonly IDataCleaningService _cleaningService;
    
    public MyService(IDataCleaningService cleaningService)
    {
        _cleaningService = cleaningService;
    }
    
    public void ProcessData(List<VitalSignData> rawData)
    {
        // 清洗数据
        var cleanedData = _cleaningService.CleanData(rawData);
        
        // 分割为窗口数据
        var windowDataList = _cleaningService.SplitIntoWindows(cleanedData);
        
        // 单独使用清洗功能
        var afterNoPeople = _cleaningService.RemoveNoPeopleSegments(rawData, thresholdMinutes: 45);
        var afterMovement = _cleaningService.RemoveLargeMovementSegments(afterNoPeople, movementThreshold: 200, afterSeconds: 60);
        var finalData = _cleaningService.FilterByMonitoringPeriod(afterMovement);
    }
}

3. 数据预处理

using MySleepStage.Core.Services;
using MySleepStage.Core.Models;

public class MyService
{
    private readonly IDataPreprocessingService _preprocessingService;
    private readonly IDataSmoothingService _smoothingService;
    
    public MyService(
        IDataPreprocessingService preprocessingService,
        IDataSmoothingService smoothingService)
    {
        _preprocessingService = preprocessingService;
        _smoothingService = smoothingService;
    }
    
    public void ProcessData()
    {
        // 预处理窗口数据
        var preprocessedWindow = _preprocessingService.Preprocess(windowData, PreprocessingMode.Auto);
        
        // 预处理单个数据数组
        double[] data = new double[] { 1.0, 2.0, 3.0, 4.0, 5.0 };
        var result = _preprocessingService.Preprocess(data, PreprocessingMode.DifferenceWithFullSmooth, _smoothingService);
        
        // 批量预处理
        var dataArray = new Dictionary<string, double[]>
        {
            { "HeartRate", heartRateData },
            { "BreathRate", breathRateData }
        };
        var batchResult = _preprocessingService.PreprocessBatch(dataArray, PreprocessingMode.Auto, _smoothingService);
        
        // 生成预处理报告
        var report = _preprocessingService.GenerateReport(originalData, preprocessedData, PreprocessingMode.Auto);
        Console.WriteLine(report);
        
        // 评估差值质量
        var assessment = _preprocessingService.AssessDifferenceQuality(differenceData);
        Console.WriteLine($"质量合格: {assessment.IsQualityAcceptable}, 推荐模式: {assessment.RecommendedMode}");
    }
}

4. 数据平滑

using MySleepStage.Core.Services;

public class MyService
{
    private readonly IDataSmoothingService _smoothingService;
    
    public MyService(IDataSmoothingService smoothingService)
    {
        _smoothingService = smoothingService;
    }
    
    public void SmoothData(double[] data)
    {
        // 三阶段组合平滑（移动平均 → 高斯滤波 → 样条插值）
        var smoothed = _smoothingService.EnhancedCombinedSmooth(data);
        
        // 单独使用各种平滑方法
        var maResult = _smoothingService.MovingAverage(data, windowSize: 12);
        var gaussianResult = _smoothingService.GaussianFilter(data, sigma: 2.5);
        var splineResult = _smoothingService.SplineSmooth(data);
        
        // 自适应阈值过滤
        var adaptiveResult = _smoothingService.AdaptiveThresholdFilter(data, sensitivityFactor: 0.7);
    }
}

5. 睡眠分期

using MySleepStage.Core.Services;
using MySleepStage.Core.Models;

public class MyService
{
    private readonly ISleepStageClassificationService _classificationService;
    
    public MyService(ISleepStageClassificationService classificationService)
    {
        _classificationService = classificationService;
    }
    
    public void AnalyzeSleep(WindowData windowData)
    {
        // 单个窗口分类
        var result = _classificationService.Classify(windowData);
        Console.WriteLine($"睡眠状态: {result.SleepState}, 置信度: {result.Confidence}");
        
        // 批量分类
        var windowDataList = new List<WindowData> { window1, window2, window3 };
        var results = _classificationService.ClassifyBatch(windowDataList);
        
        // 使用高级功能
        var standardized = _classificationService.RobustStandardize(data);
        var boundaries = _classificationService.KMeansClustering(data, clusters: 4);
        var level = _classificationService.SensitivityReducedClassify(value, boundaries, sensitivityFactor: 0.8, lastClassification: 2);
    }
}

6. 完整处理流程

public class SleepAnalysisPipeline
{
    private readonly IDataCleaningService _cleaningService;
    private readonly IDataPreprocessingService _preprocessingService;
    private readonly IDataSmoothingService _smoothingService;
    private readonly ISleepStageClassificationService _classificationService;
    
    public SleepAnalysisPipeline(
        IDataCleaningService cleaningService,
        IDataPreprocessingService preprocessingService,
        IDataSmoothingService smoothingService,
        ISleepStageClassificationService classificationService)
    {
        _cleaningService = cleaningService;
        _preprocessingService = preprocessingService;
        _smoothingService = smoothingService;
        _classificationService = classificationService;
    }
    
    public List<SleepStageResult> Analyze(List<VitalSignData> rawData)
    {
        // 1. 数据清洗
        var cleanedData = _cleaningService.CleanData(rawData);
        
        // 2. 分割窗口
        var windows = _cleaningService.SplitIntoWindows(cleanedData);
        
        // 3. 预处理每个窗口
        foreach (var window in windows)
        {
            _preprocessingService.Preprocess(window, PreprocessingMode.Auto);
        }
        
        // 4. 睡眠分期
        var results = _classificationService.ClassifyBatch(windows);
        
        return results;
    }
}

数据模型

VitalSignData

体征数据点，表示单条体征记录：

WindowData

分析窗口数据，包含指定时间窗口内的所有体征数据。

SleepState

睡眠状态枚举：

PreprocessingMode

预处理模式：

DifferenceQualityAssessment

差值序列质量评估结果：

SleepStageResult

睡眠分期结果：

配置选项

数据清洗配置

预处理配置

平滑配置

分类阈值配置

服务接口

IDataCleaningService

数据清洗服务接口：

IDataPreprocessingService

数据预处理服务接口：

IDataSmoothingService

数据平滑服务接口：

ISleepStageClassificationService

睡眠分期分类服务接口：

许可证

MIT License