What is the MeDucation Notebook?
The MeDucation Notebook is an AI study layer for hematology-oncology learners. You write notes in it, from clinic, from journal club, from a board review chapter, and the AI turns them into visual diagnostic and treatment algorithms, flashcards scheduled by the FSRS spaced repetition algorithm, board-style quizzes, dynamic mind maps, and a living knowledge graph connecting diseases, drugs, genes, biomarkers, and clinical trials. You can also chat directly with your own notes. It works alongside the MeDucation question bank or as a standalone workspace.
Written by Dr. Roupen Odabashian MD, FRCPC, FASC
Hematologist-Oncologist | Founder, MeDucation AI | Updated July 2026
Disclosure: I built this. I am the founder of MeDucation AI and a practicing hematologist-oncologist. Read this as a product explanation from someone with an obvious commercial interest, not a neutral review. There is a section below titled "When this is overkill", read it before you pay for anything.
One thing I want to be unambiguous about, because it is the whole reason the platform exists: the MeDucation question bank is reviewed by practicing hematologist-oncologist physicians, with explanations linked to PubMed. The Notebook is the AI study layer built around that human-authored content. The AI helps you organize, retrieve, and rehearse what you learn. It does not write the medicine.
What does the Notebook actually contain?
Tool | What it does | The problem it solves |
|---|---|---|
Visual algorithms | Converts your text notes into diagnostic and treatment algorithms you can see. | Your notes on, say, workup of monoclonal gammopathy are prose. The exam, and the clinic, asks you to branch. |
AI flashcards (FSRS) | Generates flashcards from your notes and schedules them with FSRS, the same algorithm Anki uses. | Making cards is the bottleneck. Most people quit spaced repetition during card creation, not during review. |
Ask My Notes | Chat with your own notes to find anything instantly. | You wrote it down six months ago. You cannot find it. Search boxes need exact words; you need the idea. |
Quiz generation | Turns your notes into board-exam-style questions. | Reading your notes feels like studying. Being asked about them is studying. |
Mind maps | Interactive maps built from your notes that auto-regenerate as you add more. | Static diagrams go stale the moment your understanding changes. |
Knowledge graph | Links diseases, drugs, genes, biomarkers, and trials across all your notebooks. | Studying in silos. You learn EGFR on Tuesday and osimertinib resistance in March and never connect them. |
Deck sharing | Share a flashcard deck with your whole program or one person by email; recipients can edit. | Every fellow in every program rebuilds the same deck from scratch, every year. |
What is a knowledge graph and why does it matter for heme/onc?
A knowledge graph is a network of nodes and connections. In the MeDucation Notebook, the nodes are the entities you actually deal with in oncology, diseases, drugs, genes, biomarkers, and clinical trials, and the edges are the relationships between them. It is built automatically as you take notes, and it spans all of your notebooks, not just the one you happen to have open. Click any node and you see its connections, a description, and which notebooks it appears in.
Here is why this matters more in our field than in almost any other. Take osimertinib. In a linear note-taking system, osimertinib lives in a page called "NSCLC, targeted therapy." In a knowledge graph, osimertinib is a node connected to EGFR, which is connected to exon 19 deletion and L858R, which are connected to osimertinib resistance mechanisms, C797S, MET amplification, small cell transformation, each of which is connected to the trials and second-line strategies you read about separately. The graph is the thing you are actually being tested on. Board questions in heme/onc are almost never "what is the mechanism of drug X." They are "this patient progressed on this drug, the next-generation sequencing shows this, what now?" That is a graph traversal question.
The graph gets denser the more you write. Eighteen months into fellowship, MYC shows up in your Burkitt notes, your double-hit lymphoma notes, and a conference trial note, and clicking that one node shows all three at once. The point is to stop studying in silos.
The evidence for this kind of visual, relational organization is real but modest. A meta-analysis of concept and knowledge maps by Nesbit and Adesope found that studying with node-link maps produced better retention and transfer than reading text or listening to lectures, though effects varied considerably by task[1]. That is a genuine effect, but I want to be careful here, because there is a well-known experiment showing that retrieval practice beat concept mapping head-to-head on a one-week retention test[2]. My read of that literature, and how I built the product: the graph is for organization and connection-finding; the flashcards and quizzes are for learning. If you only use the pretty visualizations and never test yourself, you have built a beautiful map of things you cannot recall.
How do MeDucation's mind maps work?
Mind maps are generated directly from your notes and are dynamic rather than static. As you add notes, the map auto-regenerates, so it tracks your evolving understanding instead of freezing at the version you drew in week two. You get full branch-level control: expand, collapse, rearrange, and edit individual branches.
Practically, the difference between a mind map and the knowledge graph is scope. A mind map is a hierarchy rooted in one topic, "myeloproliferative neoplasms" branching into JAK2, CALR, MPL, then into PV, ET, PMF, then into risk stratification and treatment. The knowledge graph is cross-cutting: it does not care which notebook you were in when you wrote about JAK2.
What is FSRS and why does it beat cramming?
FSRS is the Free Spaced Repetition Scheduler. It is the modern scheduling algorithm available in Anki, described in the Anki manual as "an alternative to Anki's legacy SuperMemo 2 (SM-2) algorithm" that, "by more accurately determining how much information you are likely to forget, can help you remember more material in the same amount of time"[3]. MeDucation's flashcards run on FSRS. Same algorithm; you just do not have to build the cards.
The difference from SM-2 is real. SM-2 is a hand-tuned formula from the 1980s that adjusts an "ease factor" up or down. FSRS is a machine-learning model of memory, it estimates difficulty, stability, and retrievability for each card and fits parameters to review history, letting you set a target retention rate (90% by default)[4]. In the open benchmark of spaced repetition algorithms, run on hundreds of millions of reviews from thousands of Anki users, FSRS variants consistently predict recall more accurately than SM-2[5]. Better prediction means fewer wasted reviews of cards you already know, and fewer forgotten cards you thought you knew.
Why spacing works at all is one of the oldest and best-replicated findings in cognitive psychology. Cepeda and colleagues meta-analyzed 317 experiments and found that spacing study sessions out reliably beats massing them together[6]. And retrieval, being asked, not re-reading, is the active ingredient. Roediger and Karpicke's classic demonstration of the testing effect showed that students who re-read passages outperformed testers at five minutes but were substantially worse at two days and one week[7]. That five-minute advantage is exactly the feeling of fluency you get from re-reading your notes, and it is a lie. The Dunlosky review of ten learning techniques rated practice testing and distributed practice as the only two with high utility; highlighting, re-reading, and summarization all landed in the low-utility bin[8].
Dimension | Re-reading / highlighting notes | Spaced retrieval with FSRS |
|---|---|---|
Evidence rating (Dunlosky et al.) | Low utility[8] | Practice testing + distributed practice: high utility[8] |
Short-term feeling | Feels productive; strong immediate performance[7] | Feels harder and slower |
Retention at 1 week | Falls off sharply[7] | Substantially better[7] |
Timing of review | Whenever you feel anxious about a topic | Scheduled at the point of near-forgetting; spacing beats massing[6] |
Efficiency | You re-read what is easy, not what is weak | FSRS models per-card memory state to target a chosen retention rate[4] |
What is Ask My Notes?
Ask My Notes lets you chat directly with your own notes to find anything instantly. Think of it as a personal AI assistant trained on what you wrote down, not the internet, not a textbook, your notes. "What did I write about tumor lysis prophylaxis in high-burden CLL?" It surfaces it, from whichever notebook it lives in, without you having to remember the exact phrasing you used at 6 a.m. in a workroom nine months ago.
How does deck sharing work?
You can share a flashcard deck with your whole program or with one specific person by email. Recipients can edit the deck. Importantly, an imported deck starts fresh on each person's own FSRS schedule, you inherit the cards, not my review history, which is the correct behavior, because FSRS should be modeling your memory, not mine. This is how a fellowship program builds an institutional deck across cohorts instead of every incoming fellow starting from zero.
Who is this actually useful for?
It is useful if you are early-to-mid fellowship and you are accumulating knowledge that has to survive for two or three years and then be usable in clinic afterward. It is useful if you take a lot of notes and cannot find them. It is useful if you already believe in Anki but have never had the time to make cards. It is useful for a program that wants a shared, editable deck.
When this is overkill
If you are six weeks out from the ABIM heme/onc boards, you do not need a knowledge graph. You need to grind timed, board-style questions and read the explanations. Building a beautiful interconnected map of your notes at that point is procrastination with a nice interface. The Notebook is a tool for longitudinal learning across fellowship, for the two-year accumulation, not the six-week sprint. If you are in the sprint, go do questions. (If you are earlier and want a real timeline, I wrote a 6-month study plan for the ABIM heme/onc boards.)
It is also overkill if you do not take notes. The entire system is downstream of your notes. Empty notebook, empty graph.
Frequently asked questions
Is the MeDucation question bank AI-generated?
No. The questions and explanations are written by practicing hematologist-oncologist physicians, with explanations linked to PubMed. AI is used in the Notebook, the study layer around that content, for flashcards, quizzes, mind maps, the knowledge graph, and Ask My Notes.
Is FSRS in MeDucation the same algorithm Anki uses?
Yes. FSRS (Free Spaced Repetition Scheduler) is the same algorithm available in Anki[3]. The difference is that MeDucation generates the cards from your notes, so you skip the card-authoring step that stops most people.
How much does the Notebook cost?
The Notebook, unlimited notebooks with mind maps and knowledge graph, plus AI flashcards on a spaced-repetition schedule, is part of the medical student plan, which is $18/month, or $180/year if you pay annually, which works out to $15/month. The hematology/oncology fellow plan is $225/year, which works out to $18.75/month, with shorter 3-month and 6-month billing options also available. AI usage limits apply and are listed on the pricing page, I am not going to paraphrase numbers that can change. There is a 1-week free trial on every individual plan and you can cancel any time before billing starts. Current details are on the pricing page.
Can I share flashcard decks with my fellowship program?
Yes, with your whole program or with a specific person by email. Recipients can edit the deck, and each imported copy starts fresh on that person's own FSRS schedule.
Do the mind maps update when I add new notes?
Yes. Mind maps auto-regenerate as you add notes, and you can expand, collapse, rearrange, and edit individual branches. The knowledge graph likewise grows richer and more interconnected the more you write. You can see how these fit together on the heme/onc features page.
