Features
Features:

Product Tour >

Edraw AI >

Paid Plans:

Individuals >

Business >

Eduaction >
Resources
Blog

History

How-tos & Tips

Discovery

Biography

Business Analysis

Examples

AI concept Map

Free AI Mind Map Generator

Onenote Mind Map

Bcg Matrix Examples

Nike Marketing Strategy

Unilever SWOT Analysis

Make Mind Maps in Google Docs

Guide

FAQs

What's New

Resource Center
Templates
All Templates

Brain Storming Templates

Strategy and Planning Templates

Project Management Templates

Product Management Templates

Human Resources Templates

Agile Workflow Templates

Marketing Templates

Education Templates

Fun and Games Templates

User Gallery
Download
Pricing
Enterprise

Retrospective papers

This is a mind map about a retrospective paper, which mainly introduces the relevant content and steps of writing a retrospective paper, including: submission, writing methods, statistical methods, and topic selection.

Edited at 2025-03-02 13:22:21

Mason·Carter

Recent works View more works>>

Retrospective papers

Mason·Carter

Recent works View more works>>

Recommended to you
Outline

statistics
- 19
Mason·Carter

Retrospective papers

Topic selection

Important points

novel

New technologies, new drugs, new surgeries-targeting therapeutic research

New targets, new molecules, new diagnosis-targeted diagnostic research

New models, new evaluations, new predictions-targeted predictive research

New methods, new experiments, new scales - new methods, new designs or larger-scale clinical research

feasible

Research is feasible

Is there any data for the topic selection

Is the sample sufficient?

Is the data complete?

it works

Research has clinical application value

How to choose a topic

Standing on the boss's shoulder

Utilize new surgical techniques, treatment methods and clinical drugs in the department

Standing on someone else's shoulders

Find the latest literature, understand the latest prediction models and targets, and change your thinking

eg. CNS score predicts prognosis of patients with pimcos against non-small cell lung cancer

Literature search and management

Commonly used databases

Abstract Database

English: Pubmed, Web of Science, EMBASE

Chinese: Wanfang, CNKI, VIP, CBM

Full-text database

Chinese: Full-text database of Chinese journals

English: GeenMedical (Sci-hub), ScienceDirect, SpringerLink

Search Strategy

Reading strategies

Intensive reading of literature

Comply with most keywords - full text reading - article structure, references

Rough reading of literature

Comply with a few keywords - Select to read - Method table references

Read extensive literature

Conform to one keyword - abstract reading - background discussion, reference

Read more Chinese magazine literature

Search Method

Full text search

Pubmed, Web of Science, SCI-Hub, Geenmedical recommendations, original delivery or email request

Topic selection

Magazine method: Search professional field magazines and browse the latest literature

Determine the disease method

Determined type method: Search for articles of a certain type

Comprehensive method

Check new

or and

Literature grading reading

Endnote

Article structure

Core elements

Research objects

Include exclusion criteria and clinical data

Research Methods

Intervention measures, comparison measures

Evaluation measures

Major endings, secondary endings

Statistical Methods

General statistics, special statistics

Architecture Methods

Treatment research

Target research

Statistical Methods

Data collection and entry

Research object list: baseline data, laboratory data, targets, etc.

Intervention: Treatment data

Evaluation measures: outcome indicators, follow-up results

Statistical methods and variable types

Continuous variables - input numerical value, binary classification variable - input whether (Yes=1, NO=0), multi-categorical variables - define variable names, ordered variables - level data, survival data - status time

application

Data used in all non-randomized studies/research data with confounding factors

SPSS data entry

Automatic import (recommended) - File - Import data - Excel - Select data set - Adjust variable name and variable information

Manual entry: File - New - Data - Variable View - Define variables according to Excel - Enter data

Proneness score matching PSM

concept

It means screening the experimental group and the control group through certain statistical methods, so that the selected objects are comparable in terms of certain important clinical characteristics.

principle

Usually, grouped variables are used as dependent variables, and the confounding factors that may affect the results are constructed as covariates, to obtain the tendency scores of multiple covariates of each observation model, and then match according to the proximity principle.

Testing method after PSM

Sub-theme 1

Use SPSS for PSM

Medical Statistics

Data Type

Measuring data: The measurement is the size of the value, generally there are units

Continuous variables: height, weight

Discrete variables: pregnancy, diarrhea

Description method: Normal distribution: mean ± standard deviation (discrete size); does not conform to normal distribution: median, percentile (upper and lower quartiles), median (interquartile spacing)

Counting data (categorized data): Grouped by characteristics, counting the number, generally no unit of measurement

Disordered classification data: each item is incompatible, and it is divided into two-item classification and multiple-item classification data: such as ABO blood type

Orderly classification data: All kinds of seeds have hierarchical order relationship: NYHA cardiac function grading

Description method: relative comparison, composition ratio, rate (morbidity, mortality, prevalence): number of cases (constitution ratio)

Confidence interval:

Estimate the range of possible existence of unknown overall parameters based on certain probability (confidence)

Hypothesis test

Probability of H0

Normal test method

Histogram of frequency distribution

Analysis - Description Statistics - Frequency - Chart Type) Histogram (Show normal curves in histogram—); Frequency - PP graph, Q-Q graph

KS Inspection

Nonparametric test - Old dialog - Single sample K-S, list of test variables

Exploratory test method (recommended): including histogram, Q-Q graph, K-S test

Descriptive statistics-exploration-Figure-factor-level juxtaposition, stem and leaf graph, normal graph with test

Homogeneity test of variance

Levine Inspection

For example, the homogeneity test of variance during t-test: Analysis - Comparison mean - Independent sample t-test - Select the first row result when the homogeneity of variance (p>0.05), and select the second row result when the homogeneity of variance (P<0.05).

How to deal with uneven variance

Select Nonparametric Test

Select the correct value of Brown-Forsythe or Welch

Operation: Single-factor ANOVA test-option-statistics (homogeneity test, Brown-Forsyth, Welch), case excluding according to specific analysis

When comparing two pairs, Tamhane's T2 or Dunnett's T3, etc. under the assumption of unequal variance.

Univariate ANOVA test-post hoc comparison-not assumption of equal variance (Tamheni, Dunnett T3), significant level 0.05

Measuring data inspection method

Parameter test (normal distribution)

Two groups-T tests (conditions: metrological data, single sample or two samples, normal distribution, equal variance)

Single sample t test: Comparison of whether there is any difference between the mean and the population mean of a sample

Example: Is the age of a CAD patient older than 60 years old?

Operation: Analysis - Comparison of mean - Single sample T test

Independent t test (independent sample): Comparison of whether there are differences in the mean of the two samples

Comparison of age between CABG and PCI groups

Operation: Analysis - Comparison of mean - Independent sample T test

Paired t-test (paired samples): Comparison of whether there are differences in the mean number of two paired samples

Sample: 1. Two homogeneous subjects received two different treatments; (patients of the same gender, the same condition, the same age, or paired patients after PSM; 2. Two parts of the same subject received different treatments (different drugs were given separately on the left and right arms) 3. The same subject received some treatment before and after receiving certain treatment (blood pressure before and after taking antihypertensive drugs)

Example: Comparison of age between CBG and PCI groups after PSM

Multi-group-analysis of variance (conditions: metrological data, two samples or multiple samples, normal distribution, variance)

One-way ANOVA: Comparison of whether there are differences in the mean number of samples in multiple groups

Example: Comparison of metrological data of 4 groups of patients

Operation: Analysis - Comparison of mean value - Single-factor ANOVA test - dependent variable selection measurement data to compare, factor selection group - option check description, variance homogeneity test, Brown-Forsyth, Welch)

Comparison between the mean in multiple groups: further comparison of the differences between the groups

Common tests: SNK-q test (no p-value is displayed, only if there is a statistical difference), LSD-t test (the P value is small, but it is prone to type I errors), Dunnett-t test (comparison with the control group), Bonfferoni method (conservative, available when the comparison is not many times)

Random block design (compatibility group design): Ensure that the characteristics of the observed objects in each block are as close as possible

Comparison of the measurement data of the four groups of patients, and observed whether there was a difference in the total cholesterol reduction value between the four groups and the zone groups.

Operation: Analysis - General linear model - Univariate - Select the observation indicators into the dependent variable, select the group and block group into the fixed factor - Select the construction item in the model, select the group and block group into the main effect - multiple comparisons after the event, and select the factors that need to be compared - result output, Group P < 0.01, SEX p = 0.073, indicating that there are significant differences in TOTALreduction between different groups, but no differences between different genders.

Analysis of variance of repeated measurement data: Comparison of changes in the same indicator at different time points of multiple groups of samples.

Example: Total cholesterol indexes before, after treatment, 1 month and 2 months of 4 groups of patients

Operation: Analysis - General linear model - Repeat measurement - Definition factor TIME, Number of levels 3 - Definition - Select variables at different time points - Select Group into intersubject factor - Results Analysis - Spherical test, P> 0.05, conforming to spherical distribution - TIME, P < 0.001, there are significant differences between different time points, affected by time, TIME Group, P < 0.001, there are interactions between time and group-subject effect Group, P < 0.001, there are significant differences between different groups

Nonparametric test (non-normal distribution)

Mann-Whitney U Test (Independent Sample)

Operation: Analysis - Nonparametric Test - Old Dialog Box - 2 independent samples - Select the group and comparison variables into the group variable and the test variable respectively - Options - Statistics (check description, quartiles) Exclude cases according to the test-

Wilcoxon rank sum test (paired samples)

Operation: Analysis - Non-parametric test - Old dialog box - 2 related samples - Select two paired variables in respectively - Click option, (check description, quartile) Exclude cases according to the test

Nonparametric tests for multiple independent samples

Kruskal-Wallis rank sum test

Operation: Analysis - Nonparametric Test - Old Dialog Box - K independent samples - Select group variables and test variables into the - Click options, (check description, quartiles) Exclude cases according to the test

A pair of comparisons of independent samples in multiple groups

Analysis - Nonparametric test - Independent sample - Target constant - Field (select group variables and test fields) - Settings (select custom test - Kruzkar-Wally, all in pairs) - Select Adj significance

Multiple paired samples (random zone) non-parametric test

Friedman M test (non-satisfied normal distribution and variance)

Operation - Analysis - Nonparametric Test - Old Dialog Box - K Related Samples - Select the test variables (note the grouping), test type - Fu Leedman - Statistics - Check description, quartile

Counting data inspection method

Nonparametric testing method (level data): applicable to unknown data of the overall distribution, does not conform to normal distribution, and homogeneity of variance; level data

Kruskal-Wallis rank sum test (independent sample)

By comparing the average rank of the two groups of samples, we can test whether the mean values of the two populations are significantly different. The measurement data and counting data of two independent samples are suitable for the measurement data of two independent samples.

Friedman M test (paired samples

Chi-square inspection (non-grade information)

Nonparametric test method

Survival analysis

Correlation and regression analysis

Diagnostic studies and ROC curves

Statistical inference and sample size estimation

Writing method

Submit

Free Theme

Proneness Matching Rating PSM