北京醫(yī)保藥品數(shù)據(jù)分析系統(tǒng)-系統(tǒng)介紹
本系統(tǒng)是一套基于大數(shù)據(jù)技術(shù)的北京市醫(yī)保藥品數(shù)據(jù)分析系統(tǒng)，采用Python+Django作為主要開(kāi)發(fā)語(yǔ)言和后端框架，結(jié)合Hadoop+Spark大數(shù)據(jù)處理平臺(tái)構(gòu)建完整的醫(yī)保藥品數(shù)據(jù)分析解決方案。系統(tǒng)通過(guò)集成HDFS分布式存儲(chǔ)、Spark SQL數(shù)據(jù)查詢、Pandas和NumPy科學(xué)計(jì)算庫(kù)等核心技術(shù)，實(shí)現(xiàn)對(duì)北京市醫(yī)保藥品海量數(shù)據(jù)的高效采集、存儲(chǔ)、處理和深度分析。前端采用Vue+ElementUI+Echarts技術(shù)棧打造直觀友好的用戶界面，支持藥品核心屬性分析、藥品生產(chǎn)廠家分析、藥品數(shù)據(jù)分析系統(tǒng)大屏、藥品數(shù)據(jù)挖掘分析、醫(yī)保報(bào)銷策略分析、用戶藥品數(shù)據(jù)分析大屏、用戶中藥及顆粒分析等多項(xiàng)核心功能模塊。系統(tǒng)能夠?qū)?fù)雜的醫(yī)保藥品數(shù)據(jù)通過(guò)可視化圖表形式展現(xiàn)，包括藥品價(jià)格分布、報(bào)銷比例趨勢(shì)、廠家市場(chǎng)份額、中藥使用規(guī)律等多維度分析結(jié)果，讓用戶能夠直觀地觀察和分析北京市醫(yī)保藥品的使用特征和政策效果。整個(gè)系統(tǒng)架構(gòu)采用前后端分離設(shè)計(jì)，后端通過(guò)RESTful API接口提供數(shù)據(jù)服務(wù)，前端通過(guò)Ajax異步調(diào)用獲取分析結(jié)果，確保系統(tǒng)的可擴(kuò)展性和維護(hù)性。
北京醫(yī)保藥品數(shù)據(jù)分析系統(tǒng)-選題背景
隨著我國(guó)醫(yī)療保障體系的不斷完善和醫(yī)保制度改革的深入推進(jìn)，醫(yī)保藥品數(shù)據(jù)呈現(xiàn)出數(shù)據(jù)量龐大、種類繁多、關(guān)聯(lián)關(guān)系復(fù)雜的特點(diǎn)。北京市作為全國(guó)醫(yī)療資源最為集中的地區(qū)之一，其醫(yī)保藥品使用數(shù)據(jù)具有重要的研究?jī)r(jià)值和參考意義。傳統(tǒng)的醫(yī)保數(shù)據(jù)分析方法往往依賴人工統(tǒng)計(jì)和簡(jiǎn)單的數(shù)據(jù)庫(kù)查詢，面對(duì)海量的藥品采購(gòu)記錄、報(bào)銷數(shù)據(jù)、患者用藥信息時(shí)顯得效率低下且難以發(fā)現(xiàn)深層次的數(shù)據(jù)規(guī)律。醫(yī)保管理部門和政策制定者需要通過(guò)數(shù)據(jù)分析來(lái)了解藥品使用趨勢(shì)、報(bào)銷政策效果、不同廠家藥品的市場(chǎng)表現(xiàn)等信息，以便優(yōu)化醫(yī)保政策和提高資金使用效率。同時(shí)，隨著大數(shù)據(jù)技術(shù)的快速發(fā)展，Hadoop、Spark等分布式計(jì)算框架為處理海量醫(yī)保數(shù)據(jù)提供了技術(shù)支撐，Python在數(shù)據(jù)科學(xué)領(lǐng)域的廣泛應(yīng)用也為構(gòu)建專業(yè)的醫(yī)保數(shù)據(jù)分析系統(tǒng)創(chuàng)造了條件。在這樣的背景下，開(kāi)發(fā)一套基于大數(shù)據(jù)技術(shù)的北京醫(yī)保藥品數(shù)據(jù)分析系統(tǒng)，既能滿足對(duì)醫(yī)保數(shù)據(jù)深度挖掘的實(shí)際需求，也符合當(dāng)前大數(shù)據(jù)技術(shù)在醫(yī)療健康領(lǐng)域應(yīng)用的發(fā)展趨勢(shì)。
從技術(shù)學(xué)習(xí)角度來(lái)看，本課題能夠?qū)⒋髷?shù)據(jù)理論知識(shí)與醫(yī)保數(shù)據(jù)分析的實(shí)際應(yīng)用場(chǎng)景相結(jié)合，通過(guò)構(gòu)建完整的數(shù)據(jù)處理和分析流水線，加深對(duì)Hadoop生態(tài)系統(tǒng)、Spark計(jì)算引擎、Python數(shù)據(jù)分析庫(kù)等技術(shù)的理解和運(yùn)用。項(xiàng)目涉及數(shù)據(jù)采集、清洗、存儲(chǔ)、分析、可視化等完整的技術(shù)鏈條，有助于培養(yǎng)系統(tǒng)性的大數(shù)據(jù)處理思維和解決復(fù)雜問(wèn)題的能力。從實(shí)用價(jià)值來(lái)說(shuō)，系統(tǒng)能夠?yàn)獒t(yī)保管理人員和政策研究者提供便捷的數(shù)據(jù)分析工具，幫助他們更好地理解北京市醫(yī)保藥品的使用規(guī)律和政策效果，對(duì)優(yōu)化醫(yī)保管理和政策制定具有一定的參考價(jià)值。從學(xué)習(xí)成長(zhǎng)的維度分析，項(xiàng)目整合了當(dāng)前主流的大數(shù)據(jù)處理技術(shù)和前端可視化技術(shù)，通過(guò)實(shí)際開(kāi)發(fā)過(guò)程能夠積累寶貴的工程經(jīng)驗(yàn)，為今后從事相關(guān)技術(shù)工作打下基礎(chǔ)。雖然這只是一個(gè)畢業(yè)設(shè)計(jì)項(xiàng)目，規(guī)模和復(fù)雜度相對(duì)有限，但通過(guò)認(rèn)真的設(shè)計(jì)和實(shí)現(xiàn)，依然可以在技術(shù)學(xué)習(xí)、實(shí)踐能力培養(yǎng)和解決實(shí)際問(wèn)題等方面發(fā)揮積極作用。
北京醫(yī)保藥品數(shù)據(jù)分析系統(tǒng)-技術(shù)選型
大數(shù)據(jù)框架：Hadoop+Spark（本次沒(méi)用Hive，支持定制）
開(kāi)發(fā)語(yǔ)言：Python+Java（兩個(gè)版本都支持）
后端框架：Django+Spring Boot(Spring+SpringMVC+Mybatis)（兩個(gè)版本都支持）
前端：Vue+ElementUI+Echarts+HTML+CSS+JavaScript+jQuery
詳細(xì)技術(shù)點(diǎn)：Hadoop、HDFS、Spark、Spark SQL、Pandas、NumPy
數(shù)據(jù)庫(kù)：MySQL
北京醫(yī)保藥品數(shù)據(jù)分析系統(tǒng)-圖片展示
北京醫(yī)保藥品數(shù)據(jù)分析系統(tǒng)-視頻展示
北京醫(yī)保藥品數(shù)據(jù)分析系統(tǒng)-視頻展示
北京醫(yī)保藥品數(shù)據(jù)分析系統(tǒng)-代碼展示
北京醫(yī)保藥品數(shù)據(jù)分析系統(tǒng)-代碼
from pyspark.sql import SparkSession
from django.http import JsonResponse
from django.views import View
import pandas as pd
import numpy as np
from pyspark.sql.functions import col, count, sum, avg, desc, asc, when, isnan, isnull
from pyspark.ml.feature import StringIndexer, VectorAssembler
from pyspark.ml.clustering import KMeans
import json
spark = SparkSession.builder.appName("BeijingMedicalInsuranceAnalysis").config("spark.sql.adaptive.enabled", "true").config("spark.sql.adaptive.coalescePartitions.enabled", "true").getOrCreate()
def drug_core_attribute_analysis(request):
if request.method == 'GET':
df = spark.read.format("jdbc").option("url", "jdbc:mysql://localhost:3306/medical_db").option("dbtable", "drug_info").option("user", "root").option("password", "password").load()
price_distribution = df.select("drug_price").describe().collect()
price_ranges = df.withColumn("price_range",
when(col("drug_price") < 50, "低價(jià)藥(50元以下)")
.when((col("drug_price") >= 50) & (col("drug_price") < 200), "中價(jià)藥(50-200元)")
.when((col("drug_price") >= 200) & (col("drug_price") < 500), "較高價(jià)藥(200-500元)")
.otherwise("高價(jià)藥(500元以上)")
).groupBy("price_range").agg(count("*").alias("drug_count"), avg("drug_price").alias("avg_price")).collect()
category_analysis = df.groupBy("drug_category").agg(
count("*").alias("category_count"),
avg("drug_price").alias("avg_category_price"),
sum("annual_usage").alias("total_usage")
).orderBy(desc("category_count")).collect()
dosage_form_stats = df.groupBy("dosage_form").agg(
count("*").alias("form_count"),
avg("drug_price").alias("avg_form_price")
).orderBy(desc("form_count")).collect()
prescription_type_analysis = df.groupBy("prescription_type").agg(
count("*").alias("type_count"),
avg("drug_price").alias("avg_type_price"),
sum("annual_usage").alias("type_total_usage")
).collect()
efficacy_distribution = df.groupBy("main_efficacy").agg(
count("*").alias("efficacy_count"),
avg("drug_price").alias("avg_efficacy_price")
).orderBy(desc("efficacy_count")).limit(10).collect()
price_efficacy_correlation = df.select("drug_price", "main_efficacy").toPandas()
correlation_matrix = price_efficacy_correlation.groupby('main_efficacy')['drug_price'].agg(['mean', 'count', 'std']).round(2)
price_stats = [{"stat": row['summary'], "value": float(row['drug_price'])} for row in price_distribution]
range_data = [{"range": row['price_range'], "count": int(row['drug_count']), "avg_price": round(float(row['avg_price']), 2)} for row in price_ranges]
category_data = [{"category": row['drug_category'], "count": int(row['category_count']), "avg_price": round(float(row['avg_category_price']), 2), "usage": int(row['total_usage'])} for row in category_analysis]
form_data = [{"form": row['dosage_form'], "count": int(row['form_count']), "avg_price": round(float(row['avg_form_price']), 2)} for row in dosage_form_stats]
return JsonResponse({'code': 200, 'price_stats': price_stats, 'price_ranges': range_data, 'categories': category_data, 'dosage_forms': form_data, 'message': '藥品核心屬性分析完成'})
def medical_insurance_reimbursement_analysis(request):
if request.method == 'POST':
data = json.loads(request.body)
analysis_year = data.get('year', 2023)
region = data.get('region', '全市')
df = spark.read.format("jdbc").option("url", "jdbc:mysql://localhost:3306/medical_db").option("dbtable", "reimbursement_data").option("user", "root").option("password", "password").load()
yearly_data = df.filter(col("reimb_year") == analysis_year)
if region != '全市':
yearly_data = yearly_data.filter(col("region") == region)
reimbursement_ratio_analysis = yearly_data.groupBy("drug_category").agg(
avg("reimbursement_ratio").alias("avg_reimb_ratio"),
sum("total_cost").alias("total_category_cost"),
sum("reimbursed_amount").alias("total_reimbursed"),
count("*").alias("prescription_count")
).withColumn("actual_reimb_ratio", col("total_reimbursed") / col("total_category_cost")).orderBy(desc("total_category_cost")).collect()
monthly_trends = yearly_data.groupBy("reimb_month").agg(
sum("total_cost").alias("monthly_cost"),
sum("reimbursed_amount").alias("monthly_reimbursed"),
avg("reimbursement_ratio").alias("avg_monthly_ratio")
).orderBy("reimb_month").collect()
age_group_analysis = yearly_data.withColumn("age_group",
when(col("patient_age") < 18, "未成年(18歲以下)")
.when((col("patient_age") >= 18) & (col("patient_age") < 60), "成年人(18-60歲)")
.otherwise("老年人(60歲以上)")
).groupBy("age_group").agg(
avg("reimbursement_ratio").alias("avg_age_reimb_ratio"),
sum("total_cost").alias("age_total_cost"),
count("*").alias("age_prescription_count")
).collect()
hospital_level_analysis = yearly_data.groupBy("hospital_level").agg(
avg("reimbursement_ratio").alias("avg_hospital_reimb_ratio"),
sum("total_cost").alias("hospital_total_cost"),
count("*").alias("hospital_prescription_count")
).orderBy(desc("hospital_total_cost")).collect()
policy_effectiveness = yearly_data.agg(
avg("reimbursement_ratio").alias("overall_avg_ratio"),
sum("total_cost").alias("total_medical_cost"),
sum("reimbursed_amount").alias("total_fund_expenditure")
).collect()[0]
fund_utilization_rate = float(policy_effectiveness['total_fund_expenditure']) / float(policy_effectiveness['total_medical_cost'])
category_results = [{"category": row['drug_category'], "avg_ratio": round(float(row['avg_reimb_ratio']), 3), "total_cost": float(row['total_category_cost']), "actual_ratio": round(float(row['actual_reimb_ratio']), 3)} for row in reimbursement_ratio_analysis]
monthly_results = [{"month": int(row['reimb_month']), "cost": float(row['monthly_cost']), "reimbursed": float(row['monthly_reimbursed']), "ratio": round(float(row['avg_monthly_ratio']), 3)} for row in monthly_trends]
age_results = [{"age_group": row['age_group'], "avg_ratio": round(float(row['avg_age_reimb_ratio']), 3), "total_cost": float(row['age_total_cost'])} for row in age_group_analysis]
return JsonResponse({'code': 200, 'category_analysis': category_results, 'monthly_trends': monthly_results, 'age_analysis': age_results, 'fund_utilization': round(fund_utilization_rate, 3), 'message': '醫(yī)保報(bào)銷策略分析完成'})
def drug_data_mining_analysis(request):
if request.method == 'POST':
data = json.loads(request.body)
mining_type = data.get('type', 'clustering')
df = spark.read.format("jdbc").option("url", "jdbc:mysql://localhost:3306/medical_db").option("dbtable", "drug_usage_data").option("user", "root").option("password", "password").load()
if mining_type == 'clustering':
feature_cols = ["drug_price", "annual_usage", "reimbursement_ratio", "patient_count"]
assembler = VectorAssembler(inputCols=feature_cols, outputCol="features")
feature_df = assembler.transform(df.na.fill(0))
kmeans = KMeans(k=5, seed=42, featuresCol="features", predictionCol="cluster")
model = kmeans.fit(feature_df)
clustered_df = model.transform(feature_df)
cluster_analysis = clustered_df.groupBy("cluster").agg(
count("*").alias("cluster_size"),
avg("drug_price").alias("avg_cluster_price"),
avg("annual_usage").alias("avg_cluster_usage"),
avg("reimbursement_ratio").alias("avg_cluster_reimb")
).collect()
cluster_centers = model.clusterCenters()
top_drugs_per_cluster = []
for i in range(5):
cluster_drugs = clustered_df.filter(col("cluster") == i).orderBy(desc("annual_usage")).limit(5).select("drug_name", "drug_price", "annual_usage").collect()
top_drugs_per_cluster.append({"cluster_id": i, "top_drugs": [{"name": drug['drug_name'], "price": float(drug['drug_price']), "usage": int(drug['annual_usage'])} for drug in cluster_drugs]})
elif mining_type == 'association':
frequent_combinations = df.groupBy("drug_category", "prescription_type").agg(
count("*").alias("combination_count"),
avg("drug_price").alias("avg_combination_price")
).filter(col("combination_count") > 100).orderBy(desc("combination_count")).limit(20).collect()
association_rules = []
for combo in frequent_combinations:
support = float(combo['combination_count']) / df.count()
rule_info = {
'category': combo['drug_category'],
'prescription_type': combo['prescription_type'],
'count': int(combo['combination_count']),
'support': round(support, 4),
'avg_price': round(float(combo['avg_combination_price']), 2)
}
association_rules.append(rule_info)
usage_pattern_analysis = df.groupBy("season", "drug_category").agg(
sum("annual_usage").alias("seasonal_usage"),
avg("drug_price").alias("seasonal_avg_price")
).collect()
anomaly_detection = df.filter((col("drug_price") > df.select(avg("drug_price")).collect()[0][0] + 3 * df.select(stddev("drug_price")).collect()[0][0]) | (col("annual_usage") > df.select(avg("annual_usage")).collect()[0][0] + 3 * df.select(stddev("annual_usage")).collect()[0][0])).collect()
cluster_results = [{"cluster": int(row['cluster']), "size": int(row['cluster_size']), "avg_price": round(float(row['avg_cluster_price']), 2), "avg_usage": round(float(row['avg_cluster_usage']), 2)} for row in cluster_analysis]
pattern_results = [{"season": row['season'], "category": row['drug_category'], "usage": int(row['seasonal_usage']), "avg_price": round(float(row['seasonal_avg_price']), 2)} for row in usage_pattern_analysis]
anomaly_results = [{"drug_name": anomaly['drug_name'], "price": float(anomaly['drug_price']), "usage": int(anomaly['annual_usage']), "anomaly_type": "價(jià)格異常" if anomaly['drug_price'] > 1000 else "用量異常"} for anomaly in anomaly_detection[:10]]
return JsonResponse({'code': 200, 'clusters': cluster_results, 'top_drugs_clusters': top_drugs_per_cluster, 'usage_patterns': pattern_results, 'anomalies': anomaly_results, 'message': '藥品數(shù)據(jù)挖掘分析完成'})
北京醫(yī)保藥品數(shù)據(jù)分析系統(tǒng)-文檔展示
獲取源碼-結(jié)語(yǔ)
這套基于大數(shù)據(jù)技術(shù)的北京醫(yī)保藥品數(shù)據(jù)分析系統(tǒng)算是把理論和實(shí)踐結(jié)合得比較好的一個(gè)項(xiàng)目了，從Hadoop+Spark的大數(shù)據(jù)處理到Python+Django的后端開(kāi)發(fā)，再到Vue+Echarts的數(shù)據(jù)可視化，技術(shù)棧還是挺全面的。雖然只是個(gè)畢業(yè)設(shè)計(jì)，但做下來(lái)確實(shí)能學(xué)到不少東西，特別是對(duì)醫(yī)療數(shù)據(jù)分析和大數(shù)據(jù)處理流程的理解會(huì)更深入一些。系統(tǒng)的幾個(gè)核心功能像藥品屬性分析、報(bào)銷策略分析這些，實(shí)際應(yīng)用價(jià)值也還可以。如果你也在為畢設(shè)選題發(fā)愁，或者對(duì)這個(gè)項(xiàng)目感興趣想了解更多技術(shù)細(xì)節(jié)的話，歡迎在評(píng)論區(qū)留言交流。覺(jué)得有幫助的話記得點(diǎn)個(gè)贊，需要完整資料的同學(xué)可以私信我哦！
