Cognify

Cognify is the only brain training app built on the intervention proven in a 20-year NIH trial to reduce dementia risk by 25% — adaptive visual speed training, now accessible to everyone.

Comment

Inspiration

In February 2026, Johns Hopkins and the NIH published the results of the ACTIVE study — the largest and longest cognitive training trial ever conducted. After following 2,802 adults for 20 years, they found that one specific type of brain exercise — adaptive visual speed-of-processing training — reduced dementia diagnoses by 25%. No drug, no diet, no other brain game has ever achieved this in a randomized controlled trial.

We read the paper and had the same thought: if this works, why isn't it in everyone's hands? BrainHQ licenses the original exercise but charges $14/month and buries it inside 29 other unproven games. Lumosity was fined $2M by the FTC for making claims they couldn't back up. There was no clean, focused, accessible app that delivered the exact intervention the science validated. So we built one.

What it does

Cognify delivers the specific adaptive visual speed-of-processing protocol from the ACTIVE study as a free, accessible web app designed for older adults. It includes:

  • Three progressive exercises — Central Identification, Divided Attention, and Selective Attention — each modeled directly on the phases described in the ACTIVE study
  • The exact adaptive algorithm — a 3-up/1-down staircase procedure that converges on the user's ~79% accuracy threshold, the optimal difficulty for driving neuroplastic change
  • A full onboarding flow with baseline assessment, age-group comparison, and recommended training schedule
  • Personalized progress tracking — processing speed trends, accuracy charts, weekly session frequency, milestone badges, and research-backed estimates of dementia risk reduction based on training progress
  • Cited research throughout — every statistic shown in the app is sourced from a named study (NIH, Cambridge Cam-CAN, Karolinska, WHO, etc.), displayed inline during training breaks and on the dashboard
  • Regulatory compliance — Terms & Conditions, Privacy Policy, and Data Consent screens with honest medical disclaimers
  • An account system with SHA-256 password hashing via the Web Crypto API and persistent cross-session data storage

The target audience is adults 50+. Every design decision — large text, high contrast, simple flows, breathing breaks between blocks, research-backed encouragement — is made with senior usability in mind.

How we built it

Cognify is a single-page React application with no backend. The entire app lives in one JSX file using:

  • React with hooks (useState, useEffect, useRef, useCallback, useContext, useMemo) for state management
  • CSS custom properties for a complete light/dark theme system — theme switching happens via a data-theme attribute on the document root with zero React re-renders
  • requestAnimationFrame for frame-accurate stimulus timing (critical — the entire training depends on displaying stimuli for precise durations down to ~17ms)
  • Recharts for progress visualization (processing speed trends, accuracy, weekly session frequency)
  • Web Crypto API (SHA-256) for secure password hashing
  • Persistent storage API for cross-session data (user accounts, training history, settings)
  • Responsive CSS with 4 breakpoints scaling from 375px phones to ultrawide monitors, plus 100dvh and env(safe-area-inset-bottom) for mobile browser chrome handling

No backend, no database, no build tools beyond what the artifact environment provides. All data is client-side.

Challenges we ran into

  • Timing precision is the hardest technical problem. The training depends on showing a shape for exactly X milliseconds — but browsers don't give you precise timing with setTimeout. We use requestAnimationFrame and count frames (each ~16.67ms at 60fps), which means our minimum resolution is one frame. The original ACTIVE study used CRT monitors with higher refresh rates; we acknowledge this as a limitation.
  • Designing for seniors forced us to rethink every default. Font sizes that feel normal to us (14–16px) are too small. Tap targets that seem fine on our phones are too small. We ended up with 18px body text, 20px buttons, and 48px icon targets — and it still felt tight on testing.
  • Balancing motivation with honesty. We wanted to show users their estimated dementia risk reduction, but we had to be careful not to make medical claims. Every statistic is framed as "based on the ACTIVE study" or "estimated from population-level data," never as a personal medical prediction. The legal disclaimers are real and thorough.
  • Keeping it focused. The temptation to add more exercises, social features, gamification layers, and customization is strong. We kept returning to the research: only speed-of-processing training worked. Only adaptive difficulty mattered. Only boosters made it last. The app does one thing — the thing that was proven — and nothing else.

Accomplishments that we're proud of

  • The adaptive staircase algorithm matches the published protocol exactly — 3-up/1-down, asymmetric step sizes, converging on the ~79% accuracy threshold
  • 14 real research citations from named institutions woven throughout the user experience, not hidden in a footnote
  • A complete regulatory compliance flow (Terms, Privacy, Data Consent) that would hold up to real scrutiny
  • An app that our parents and grandparents could actually use without asking for help

What we learned

  • The gap between "published science" and "accessible product" is enormous. The ACTIVE study has been public since February 2026, but turning its protocol into something a 75-year-old can use on their iPad requires hundreds of design decisions the paper never addresses.
  • Senior-friendly design isn't just "make it bigger." It's about reducing cognitive load at every step — fewer choices per screen, clearer feedback, explicit encouragement, and never making the user feel like they failed.
  • The strongest competitive moat isn't technology — it's being the only product that can honestly cite a specific, named, 20-year RCT. Every other brain training app is making claims they can't back up.

What's next for Cognify

  • Native iOS and Android apps for better timing precision and push notification booster reminders
  • A pilot study partnering with a university lab to validate our implementation against the original ACTIVE protocol
  • FDA 510(k) clearance as a Class II digital therapeutic (BrainHQ's Double Decision has precedent)
  • Institutional licensing for senior living facilities and memory care clinics
  • An anonymized research database contributing to the next generation of cognitive training science ```

By: Enrique Reid, Jasper Gilley, Khushaan Virk, Dean Kiyingi

Built With

Share this project:

Updates