AI for Biometrics

Study Lifecycle Monitor

A visual, runnable guide to the operational-health console

Run / share / read / build - pick your mode. This page shows the console (diagrams + a live roll-up computed from the portfolio), the run-and-build page lets you run it and nudge a what-if, the dashboard is a shareable snapshot, and the run-and-build page shows how it is made. Synthetic, no PHI, no participant-level data. A screening lens for operational health, never a reported clinical number.

What it watches

Operational health of every study from protocol to CSR, on one screen: are milestones slipping, is the query backlog clearing before database lock, is TLF production keeping pace, is double-programming on track.

%%{init: {'theme':'base','themeVariables':{'primaryColor':'#28344f','primaryTextColor':'#eef2fb','primaryBorderColor':'#7d9bd6','lineColor':'#aab8d6','clusterBkg':'#161d33','clusterBorder':'#3a476a','tertiaryColor':'#20406a','tertiaryTextColor':'#eef2fb','tertiaryBorderColor':'#7d9bd6','secondaryColor':'#20406a','secondaryTextColor':'#eef2fb'}}}%%
flowchart LR
  S[Startup] --> C[Conduct] --> L[DB Lock] --> R[Reporting]
  S -.-> M1["Protocol<br/>SAP · FPFV"]
  C -.-> M2[LPLV]
  L -.-> M3[DBL]
  R -.-> M4["TLFs · CSR"]

How the risk score works - it is transparent

No black box. Each study gets four operational sub-scores (0-100), blended into one:

%%{init: {'theme':'base','themeVariables':{'primaryColor':'#28344f','primaryTextColor':'#eef2fb','primaryBorderColor':'#7d9bd6','lineColor':'#aab8d6','clusterBkg':'#161d33','clusterBorder':'#3a476a','tertiaryColor':'#20406a','tertiaryTextColor':'#eef2fb','tertiaryBorderColor':'#7d9bd6','secondaryColor':'#20406a','secondaryTextColor':'#eef2fb'}}}%%
flowchart LR
  T["Timeline<br/>overdue + lock pressure"] --> O
  D["Data<br/>queries + clean/SDV"] --> O
  F["TLF<br/>finalized vs pressure"] --> O
  Rs["Resource<br/>double-prog + discrep"] --> O
  O{{"overall = 0.30·T + 0.28·D<br/>+ 0.27·F + 0.15·R"}}
  O --> RAG["RAG bands<br/>red · amber · green"]

Because it is a formula, you can defend every number and change what “risk” means in one place (score_study() in R/logic.R).

The live portfolio - computed from the data

Everything below is rendered from four CSVs, the same files the Shiny app reads: studies.csv, milestones.csv, deliverables.csv, and query_trend.csv (in the shiny/data/ folder). Edit those CSVs (or run the app) and these move.

8

Active studies

1

At risk

3

On watch

148

Open queries

63%

Avg deliverables

33d

Next DB lock

Severity	Study	Signal	Detail
red	CP-108	TLF production behind	10% finalized while in Reporting
red	CP-108	Milestone overdue	TLFs, CSR past due
red	CP-106	TLF production behind	34% finalized while in DB Lock
red	CP-105	TLF production behind	24% finalized while in DB Lock
amber	CP-108	Double-programming lag	42% double-programmed, 16 open discrepancies
amber	CP-106	Double-programming lag	58% double-programmed, 14 open discrepancies
amber	CP-105	Double-programming lag	43% double-programmed, 2 open discrepancies
amber	CP-107	TLF production behind	44% finalized while in Reporting
amber	CP-107	Double-programming lag	55% double-programmed, 7 open discrepancies

Run it - and change it

shiny::runApp("lifecycle_dashboard/shiny")
# needs: install.packages(c("shiny","bslib","DT","ggplot2"))

Pick a study, then move the what-if sliders (open queries, double-programming) and watch that study’s row in the heatmap re-colour. The run-and-build page shows how the app is built from R/logic.R.

Going live: R/gen_data.R writes the four CSVs (studies.csv, milestones.csv, deliverables.csv, query_trend.csv) so you can see the shape. Replace them with a read-only CTMS / EDC export carrying the same columns and the console is your portfolio. Nothing writes back to a source system.

The other formats

Mode	Artifact
Run	Run & build the Shiny app
Share (no R)	Dashboard
Build	Run-and-build walkthrough

Synthetic, deterministic, no PHI. Operational screening, not a clinical readout.

--- title: "Study Lifecycle Monitor" subtitle: "A visual, runnable guide to the operational-health console" format: html: theme: - darkly - ../../_quarto-mc.scss include-before-body: ../../_mc-topbar.html toc: true toc-location: left toc-title: On this page embed-resources: true code-tools: true code-fold: true execute: echo: false warning: false message: false knitr: opts_chunk: dev: png --- ```{r setup} library(ggplot2); library(knitr) source("../shiny/R/logic.R") d <- lifecycle_load("../shiny/data") st <- score_portfolio(d) k <- lifecycle_kpis(st, d) ``` ```{=html} <style> svg[id^="mermaid"]{ width:100% !important; max-width:940px; height:auto; display:block; margin:8px auto; } svg[id^="mermaid"] .nodeLabel, svg[id^="mermaid"] .edgeLabel { font-size:16px !important; } </style> ``` ::: {.callout-note appearance="simple"} **Run / share / read / build - pick your mode.** This page *shows* the console (diagrams + a live roll-up computed from the portfolio), the [**run-and-build page**](lifecycle-build.html) lets you *run* it and nudge a what-if, the [dashboard](lifecycle-dashboard.html) is a shareable snapshot, and the [run-and-build page](lifecycle-build.html) shows how it is made. Synthetic, no PHI, no participant-level data. A screening lens for operational health, never a reported clinical number. ::: ## What it watches Operational health of every study from protocol to CSR, on one screen: are milestones slipping, is the query backlog clearing before database lock, is TLF production keeping pace, is double-programming on track. ```{mermaid} %%{init: {'theme':'base','themeVariables':{'primaryColor':'#28344f','primaryTextColor':'#eef2fb','primaryBorderColor':'#7d9bd6','lineColor':'#aab8d6','clusterBkg':'#161d33','clusterBorder':'#3a476a','tertiaryColor':'#20406a','tertiaryTextColor':'#eef2fb','tertiaryBorderColor':'#7d9bd6','secondaryColor':'#20406a','secondaryTextColor':'#eef2fb'}}}%% flowchart LR S[Startup] --> C[Conduct] --> L[DB Lock] --> R[Reporting] S -.-> M1["Protocol<br/>SAP · FPFV"] C -.-> M2[LPLV] L -.-> M3[DBL] R -.-> M4["TLFs · CSR"] ``` ## How the risk score works - it is transparent No black box. Each study gets four operational sub-scores (0-100), blended into one: ```{mermaid} %%{init: {'theme':'base','themeVariables':{'primaryColor':'#28344f','primaryTextColor':'#eef2fb','primaryBorderColor':'#7d9bd6','lineColor':'#aab8d6','clusterBkg':'#161d33','clusterBorder':'#3a476a','tertiaryColor':'#20406a','tertiaryTextColor':'#eef2fb','tertiaryBorderColor':'#7d9bd6','secondaryColor':'#20406a','secondaryTextColor':'#eef2fb'}}}%% flowchart LR T["Timeline<br/>overdue + lock pressure"] --> O D["Data<br/>queries + clean/SDV"] --> O F["TLF<br/>finalized vs pressure"] --> O Rs["Resource<br/>double-prog + discrep"] --> O O{{"overall = 0.30·T + 0.28·D<br/>+ 0.27·F + 0.15·R"}} O --> RAG["RAG bands<br/>red · amber · green"] ``` Because it is a formula, you can defend every number and change what "risk" means in one place (`score_study()` in `R/logic.R`). ## The live portfolio - computed from the data {#live} Everything below is rendered from four CSVs, the same files the Shiny app reads: [`studies.csv`](../shiny/data/studies.csv), [`milestones.csv`](../shiny/data/milestones.csv), [`deliverables.csv`](../shiny/data/deliverables.csv), and [`query_trend.csv`](../shiny/data/query_trend.csv) (in the [`shiny/data/`](../shiny/data/) folder). Edit those CSVs (or run the app) and these move. ```{r kpis, results='asis'} box <- function(label, value, col) sprintf( '<div style="flex:1 1 120px;background:#222630;border:1px solid #3a4157;border-radius:10px;padding:14px 16px"><div style="font-size:28px;font-weight:700;color:%s;font-family:monospace;line-height:1">%s</div><div style="font-size:11px;color:#94a1c0;text-transform:uppercase;letter-spacing:.05em;margin-top:6px">%s</div></div>', col, value, label) cat('<div style="display:flex;flex-wrap:wrap;gap:10px;margin:6px 0 18px">') cat(box("Active studies", k$n, "#e9edf7")) cat(box("At risk", k$at_risk, "#e8657a")) cat(box("On watch", k$watch, "#e0a64c")) cat(box("Open queries", k$open_q, "#e9edf7")) cat(box("Avg deliverables", paste0(k$avg_del,"%"), "#3fd0c9")) cat(box("Next DB lock", ifelse(is.na(k$next_dbl_days),"-",paste0(k$next_dbl_days,"d")), "#5b8cff")) cat('</div>') ``` ::: {.panel-tabset} ### Risk heatmap ```{r heat, fig.width=8, fig.height=4} dims <- c("Timeline","Data","TLF","Resource") hl <- do.call(rbind, lapply(dims, function(dm) data.frame(id=st$id, dim=dm, value=st[[dm]]))) hl$id <- factor(hl$id, levels=rev(st$id)); hl$dim <- factor(hl$dim, levels=dims) ggplot(hl, aes(dim, id, fill=value)) + geom_tile(colour="#222630", linewidth=1.5) + geom_text(aes(label=value), colour="#0b1020", fontface="bold", size=4) + scale_fill_gradientn(colours=c("#43c47d","#e0a64c","#e8657a"), limits=c(0,100)) + labs(x=NULL, y=NULL) + theme_minimal(base_size=12) + theme(legend.position="none", panel.grid=element_blank(), plot.background=element_rect(fill="#222630", colour=NA), panel.background=element_rect(fill="#222630", colour=NA), text=element_text(colour="#e9edf7"), axis.text=element_text(colour="#c5cee4")) ``` ### Ranked signals ```{r sig} sg <- lifecycle_signals(st, d) kable(head(sg[, c("sev","who","title","msg")], 10), col.names=c("Severity","Study","Signal","Detail"), row.names=FALSE) ``` ::: ## Run it - and change it {#run} ```r shiny::runApp("lifecycle_dashboard/shiny") # needs: install.packages(c("shiny","bslib","DT","ggplot2")) ``` Pick a study, then move the what-if sliders (open queries, double-programming) and watch that study's row in the heatmap re-colour. The [run-and-build page](lifecycle-build.html) shows how the app is built from `R/logic.R`. ::: {.callout-tip appearance="simple"} **Going live:** `R/gen_data.R` writes the four CSVs ([`studies.csv`](../shiny/data/studies.csv), [`milestones.csv`](../shiny/data/milestones.csv), [`deliverables.csv`](../shiny/data/deliverables.csv), [`query_trend.csv`](../shiny/data/query_trend.csv)) so you can see the shape. Replace them with a read-only CTMS / EDC export carrying the same columns and the console is your portfolio. Nothing writes back to a source system. ::: ## The other formats | Mode | Artifact | |---|---| | Run | [Run & build the Shiny app](lifecycle-build.html) | | Share (no R) | [Dashboard](lifecycle-dashboard.html) | | Build | [Run-and-build walkthrough](lifecycle-build.html) | *Synthetic, deterministic, no PHI. Operational screening, not a clinical readout.*