Data and code from: A meta-analysis suggests that TMS targeting the hippocampal network selectively improves episodic memory
Data files
Mar 24, 2026 version files 96.15 KB
-
analysis_dataset.csv
45.20 KB
-
code.zip
34.17 KB
-
HeaderDataDictionary_v3.txt
10.92 KB
-
README.md
5.87 KB
Abstract
Episodic memory is critically dependent on the hippocampal network and is frequently impaired in many clinical disorders. Recent findings highlight Hippocampal Indirectly Targeted Stimulation (HITS) as a promising, noninvasive transcranial magnetic stimulation (TMS) procedure to enhance episodic memory performance. Here, we report the first comprehensive meta-analysis of HITS effects on episodic memory, encompassing both healthy individuals and clinical populations. HITS robustly improved episodic memory, with effects selective for episodic memory versus other non-memory cognitive domains. Efficacy was significantly greater when memory performance was assessed using memory tasks sensitive to recollection, which is strongly linked to hippocampal network function, compared to recognition or other types of episodic memory tasks. Efficacy was also significantly greater when HITS was delivered before the memory tasks were administered, versus in the period between the study and test phases of tasks. No serious adverse events were reported. These findings establish HITS as a robust approach for episodic memory enhancement, suggesting potential for clinical translation in memory disorders. Selectivity of effects for episodic memory generally and for recollection-format tests in particular indicates cognitive and mechanistic specificity, supporting the potential for targeted and selective neuromodulation of hippocampal networks and their associated functions.
Dataset DOI: 10.5061/dryad.vhhmgqp86
Description of the data and file structure
The analysis_dataset.csv file is a table of standardized effect sizes used in the meta-analysis, organized by reviewed study and categorized based on the experiment design factors that were assessed. The data are organized such that each row has one effect as well as all the study and factor categorizations for that effect, organized by columns. The first row is a header that indicates the study and factor information provided in each column.
The HeaderDataDictionary_v3.txt file is a description of how the data in analysis_dataset.csv correspond to the factor descriptions used in the meta-analysis reported in the corresponding publication, described in Tables 1 and 2 of the HeaderDataDictionary_v3.txt file (and in Table 1 and Table S1 in the corresponding publication). Each factor in Table 2 is mapped onto a corresponding column label/header in analysis_dataset.csv, and the different levels of each factor are mapped onto the corresponding labels in analysis_dataset.csv. Some columns in analysis_dataset.csv contain information that was not included in the meta-analysis (e.g., the names of the different tasks that were used). All columns that were not included in the meta-analysis are indicated in their descriptions in HeaderDataDictionary.txt.
Code.zip is a repository of the R code needed to reproduce all analyses and figures in the corresponding publication. It includes a README.md file that describes the organization of the data and code, the required R version and packages needed to execute the code, and instructions for executing the code ("run_all.R").
Files and variables
File: analysis_dataset.csv
Variables
- StudyNumber: Corresponds to study numbers provided in Table 1
- StudyName: First author and publication year of corresponding study
- N: Sample size for the effect described in this row
- OUTCOME_TYPE: Type of task used 'Mem'=episodic memory, 'Other'=non-memory
- TaskType: Corresponds to the "Task" factor in Table S1. For episodic memory tasks, those that measured recollection versus recognition versus other. 'NA' indicates a non-memory task
- TaskName: Verbal description of the type of task in the study (not analyzed)
- Target: Corresponds to the "Targeting" factor in Table S1. Individual='individual'. Group/atlas='group'.
- Timing: Corresponds to the "HITS Timing" factor in Table S1. Pre-task='before'. Post-encoding='after'.
- Sample: Corresponds to the "Population" factor in Table S1. Younger adult='YA'. Older adult='OA'. MCI/AD='MCI_AD', Other='Other'.
- StimLocation: Corresponds to the "Target" factor in Table S1. Left parietal='LPC'. Other='Other'.
- TestDelay: Corresponds to the "Task delay" factor in Table S1. Short='short'. Medium='med'. Long='long'.
- SessionBin: Corresponds to the "Sessions" factor in Table S1. 1='1'. 2-4='2-5'. 5+='>5'
- nSession: The number of stimulation sessions, not binned according to the ranges in SessionBin (not analyzed).
- IntensityBin: Corresponds to the "Intensity" factor in Table S1. 100+ MT='>=100'. <100% MT='<100'.
- Intensity: Further description of the IntensityBin factor for each study. R is for resting motor threshold. A is for active motor threshold. The number indicates the % of the corresponding motor threshold.
- ControlBin: Corresponds to the "Control" factor in Table S1. Baseline='Baseline'. Other location='Control'. Low intensity='LowInt'. Other='Other'.
- ComparisonBin: Corresponds to the "Design" factor in Table S1. Post-Post = 'post_stim_sham'. Pre-Post = 'pre_post_stim'. Delta = 'stim_sham_delta'. Interaction = 'time_stim_int'. Other = 'other'.
- ComparisonType: Further description of the Other category for the ComparisonBin factor (not analyzed).
- TMSProtocolBin: Corresponds to the "TMS protocol" factor in Table S1. rTMS='TMS'. TBS='TBS'.
- TMSProtocol: Further description of the TMSProtocolBin factor (not analyzed). Indicates the frequency for TMS, whether TBS was intermittent or continuous ('i' or 'c'), and the burst frequency for TBS.
- EffectSize: Reported or computed effect size, either Cohen's d or partial-eta-squared.
- g: Computed Hedges' g statistic.
- var_g: variance of the computed g statistic
- Outlier: whether the effect size estimate was treated as an outlier for the sensitivity analysis.
File: HeaderDataDictionary.txt
Description: This text file lists the same information provided above, but with more detail.
Code/software
The code is described in the "README.md" file within code.zip. The code operates on a copy of the "analysis_dataset.csv" data file, described above. The copy is labeled "analysis_dataset_Final.csv", with all variables identical to "analysis_dataset.csv".
R version 4.2.0 or greater is required. Required packages include:
"metafor", # Meta-analysis models
"glmnet", # LASSO regularization
"dplyr", # Data manipulation
"tidyr", # Data reshaping
"stringr", # String processing
"readr", # File I/O
"ggplot2", # Visualization
"patchwork", # Plot composition
"gt", # Publication tables
"ggsci", # Scientific journal color palettes
"data.table", # Fast data reshaping
"RColorBrewer", # Color palettes
"scales", # Axis formatting
"ggrepel", # Non-overlapping text labels
"gridExtra", # Grid-based plot arrangement
"corrplot", # Correlation matrices
"ggcorrplot", # ggplot2 correlation matrices
"ggalluvial", # Alluvial/Sankey diagrams
"extrafont", # Font support (optional)
"marginaleffects" # Marginal predictions
Access information
Other publicly accessible locations of the data:
- None
Data were derived from the following sources:
- Systematic review as described in the manuscript
