Now in open beta - v0.4

Research analytics, finally reimagined

Alphensi is the analytical studio that replaces jamovi, JASP, and scattered R scripts. With a reactive, reproducible, visually stunning environment built for serious researchers.

Early Access → The difference

24+

Analysis types

100%

Open source core

R 4.5

Runtime engine

APA 7

Auto-formatted

alphensi.app / project / clinical-study-alpha

Alphensi v0.4

File Analysis Visualise Report clinical_study_alpha.csv

R 4.5.1 ready

Analyses

📊 Descriptives

Descriptive Stats

⚖️ T-Tests

📈 ANOVA

One-Way ANOVA

〰️ Regression

Linear Regression

Logistic Regression

Mixed Models

🎲 Bayesian

⏱️ Survival

Kaplan–Meier

Variables

numage

numbmi

numcholesterol

cattreatment

binevent

⊞ Data

◎ Results

◈ Plots

⟳ Pipeline

</> R Code

📄 Report

Model Fit - OLS Regression: cholesterol ~ age + bmi + systolic_bp

copyWord

0.41

R²

0.38

Adj. R²

14.7

F-stat

< .001

Model p

Predictor	B	β	t	p	95% CI
(Intercept)	42.3	—	—	—	—
age	0.88	0.31	4.2	< .001	[0.47, 1.29]
bmi	2.14	0.28	3.8	.002	[1.03, 3.25]
systolic_bp	0.42	0.14	2.1	.038	[0.02, 0.82]

The OLS model explained 41% of variance in cholesterol. Significant predictors: age (β=0.31), bmi (β=0.28), systolic_bp (β=0.14).

Residuals vs Fitted - assumption check

⬇ PNG

n = 247 missing = 4 (1.6%) analysis = Linear Regression clinical_study_alpha.csv · snapshot a3f8c21d

Linear Regression

OLS with diagnostics & CIs

Options

Plots

Assumptions

Outcome (Y)

cholesterol ✕

Predictors (X)

age ✕ bmi ✕ systolic_bp ✕

Coefficients, SE, t, p Standardised β R², Adj R², F-test 95% CIs VIF collinearity

▶ Run Analysis

The problem

Existing tools are holding your research back

jamovi and JASP made statistics accessible. But the paradigm, static menus, isolated output blocks, manual reproducibility, hasn't kept pace with modern research demands.

✗

Static, one-block-at-a-time outputs

Run an analysis, get a frozen table. Change a variable, start over. No reactivity, no live updates, no undo.

✗

Visualisations you'd be embarrassed to publish

Basic bar charts with no interactivity. No Plotly, no effect plots, no posterior distributions, no SHAP panels.

✗

Reproducibility as an afterthought

Analyse in the GUI, copy-paste to Word, lose the trail. Three colleagues, three different results from the same dataset.

✗

Advanced models require writing code anyway

Bayesian hierarchical models, mixed effects, survival regression, all fall back to "write R yourself" the moment complexity increases.

The shift

jamovi / JASP / RStudio Old paradigm

○Sequential dialogs, click-to-run

○Static APA tables, no interactivity

○Manual export to Word/PDF

○GUI mode OR code mode, never both

○Version control requires external Git

Alphensi New paradigm

●Reactive analytical graph, everything live-updates

●Plotly + D3 interactive charts, effect plots, posterior viz

●Auto-generated Quarto document, one-click PDF/Word

●GUI actions write R code in real time, switch anytime

●Version history, snapshot hashes, rollback built in

📡 Analytical engine

A reactive graph, not a sequence of dialogs

Every dataset, transformation, and model is a node in a persistent analytical graph. Change a filter upstream, every downstream model, chart, and table updates instantly. No re-running. No stale results.

Full dependency tracking across all objects

Undo/redo across the entire analysis chain

Branching, compare two analytical paths side by side

Snapshot hashes guarantee exact reproducibility

Analytical Graph · clinical_study_alpha

📊 Data Object

247 rows · 14 vars · SHA: a3f8c21d

⟳ Pipeline

→

filter · recode · derive ✓ 3 active

GLM · cholesterol

KM · time_to_event

PCA · 6 vars

◈ Residual plot

◈ KM curves

◈ Scree + Biplot

📄 Quarto .qmd

● Upstream change detected - 3 models queued for update

</> Code generation

Every GUI action writes real R code

The Code tab is never empty. Every variable assignment, checkbox tick, and method selection immediately produces valid, runnable tidyverse R. Switch to code mode at any moment — edit freely, and the GUI reflects your changes.

Tidyverse + rstatix + broom + lme4 output

One-click copy, or download the full project script

No "GUI mode vs code mode", they're the same state

Syntax-highlighted with package attribution

</> R Code, auto-generated

</> R Code

📄 Report

copy ⬇ .R

# ── Kaplan–Meier Survival ── Alphensi v0.4 # Project: clinical_study_alpha.csv # Generated: 2025-04-06 14:22:07 library(tidyverse); library(survival) library(survminer) # Pipeline df <- read_csv("clinical_study_alpha.csv") |> filter(age > 18, !is.na(bmi)) |> mutate(bmi_cat = case_when( bmi < 25 ~ "Normal", bmi < 30 ~ "Overweight", TRUE ~ "Obese")) # KM fit km_fit <- survfit( Surv(time_to_event, event) ~ treatment, data = df ) ggsurvplot(km_fit, data = df, pval = TRUE, risk.table = TRUE) survdiff(Surv(time_to_event, event) ~ treatment, data = df) # log-rank

📊 Advanced modelling

Advanced models with full GUI support

Logistic regression with ROC/AUC, Kaplan–Meier survival curves, PCA with biplot, Linear Mixed Effects with ICC, all first-class. No R required. Every model includes diagnostics, effect plots, and an auto-written interpretation paragraph.

Logistic regression - OR forest plot, ROC curve, calibration

Kaplan–Meier - log-rank test, median survival, risk table

PCA — scree plot, loadings heatmap, PC1×PC2 biplot

Mixed effects - ICC, marginal/conditional R², BLUP caterpillar

Bayesian T-Test - BF₁₀, prior vs posterior, robustness

Logistic Regression - OR Forest Plot

0.79

AUC

0.28

Nagelkerke R²

.312

H-L p

Events

Odds ratios with 95% confidence intervals

age

1.82

<.001

bmi

1.45

.003

systolic_bp

1.18

.042

sex (Male)

1.07

.38

hba1c

0.84

.21

OR = 1.0 (null) marked. Error bars = 95% CI. ● = point estimate.

📄 Reproducibility

From GUI to published paper in one click

Reproducibility isn't a feature you enable, it's the default. Every project stores dataset hashes, package versions, transformation steps, and model specs. A complete Quarto document is auto-generated and always current.

Live Quarto .qmd auto-written from the analytical graph

One-click render to PDF (APA 7), Word, or interactive HTML

Full manifest: SHA-256 dataset hash, R version, all package versions

Version history with rollback, git for your analysis

Collaborative annotation: colleagues extend, not overwrite

📄 Report - Auto-generated Quarto document

Auto-generated Quarto document

live PDF Word HTML

--- title: "Statistical Analysis Report" date: "6 April 2025" format: { pdf: {}, html: { toc: true }, docx: {} } execute: { echo: false, warning: false } ---

Statistical Analysis Report

1. Dataset & Pipeline

N = 247, 14 variables. Pipeline: 3 active steps. Missing: 4 cells (1.6%), listwise deletion.

```{r setup} df <- read_csv("clinical_study_alpha.csv") |> filter(age > 18, !is.na(bmi))```

2. Kaplan–Meier Survival Analysis

Log-rank test: χ²(2) = 7.4, p = .025 - groups differed significantly in survival. Median: Control = 28 months; Treatment A = 44 months.

4. Reproducibility Manifest

SHA: a3f8c21d R 4.5.1 tidyverse 2.0 survival 3.5 Alphensi 0.4.0

⬇ Download .qmd

PDF / APA 7

Word

24+ analyses, all first-class

Every method ships with full GUI configuration, interactive plots, assumption checks, APA output, and live R code.

📊

Descriptives

Full summary stats, Shapiro-Wilk, outlier detection, distribution explorer.

meanSDnormality

⚖️

T-Tests

Independent, paired, one-sample. Welch correction, Cohen's d, Bayes factor.

Cohen's dCILevene

📈

ANOVA

One-way, two-way, repeated measures, ANCOVA. η², ω², Tukey HSD post-hoc.

η²TukeyMauchly

〰️

Regression

OLS, logistic, GLM, mixed effects. Diagnostics, VIF, standardised β, OR plots.

R²ORlme4

🔗

Correlations

Pearson, Spearman, partial correlations. Interactive heatmap with significance overlay.

heatmappartial

⏱️

Survival

Kaplan–Meier curves, log-rank test, Cox proportional hazards, risk table.

KMCox HRlog-rank

🎲

Bayesian

T-test, regression, hierarchical (brms). BF₁₀, posterior distribution, robustness check.

BF₁₀brmsStan

🔲

Factor Analysis

PCA with scree + biplot, EFA with rotation methods, CFA via lavaan.

PCAEFACFA