Dorus Reading List
The shortest path into a new topic.
Describe your project. Get what to read, in what order, and a self-test that checks it stuck.
Citation intent
Paper type & difficulty
Open-access copies
Takeaways with quotes
I'm starting a project on detecting atrial fibrillation from a smart mattress.⏎
Background: signal processing
Time: 4 weeks
Level: graduate
Papers in the path
8
3 classics from before the atlas window
Reading time
5h 15m
About four evenings
Free to read
8 of 8
1 via a preprint copy
Self-test
12 questions
Why and how, each with its quote
Foundations
- 1Ballistocardiography: a tutorial on the mechanical signature of the heartbeatTake from it: what one BCG beat looks like, and why posture changes it.
- 2Atrial fibrillation for engineers: rhythm, burden and diagnosisTake from it: why irregular beat intervals are the signal, and what a Holter counts as an episode.
- 3Beat-to-beat interval detection from bed sensorsTake from it: the J-peak detector that most later work extends.
Methods
- 4Irregularity features for AF detection from inter-beat intervalsTake from it: window statistics that survive a few missed beats.
- 5An open overnight BCG dataset with synchronized ECGTake from it: the benchmark most results use, and why its subject-level folds matter.
State of the art
- 6Ballistocardiographic AF detection in a sleep laboratoryTake from it: the strongest lab result, and what changes outside the lab.
- 7Contactless AF screening with an under-mattress sensor: a 120-night home studyTake from it: the current home benchmark. Read its claims with the Paper Report open.
Open problems
- 8Short AF episodes from contactless sensors: where current methods failTake from it: three open questions you could build a project on.
Why this orderfrom citation intent
Atlas paper
Classic, found via citations
An arrow means "cited as background". #4 cites #2 and #3 that way, so both come first.
Left out, on purpose
- 31Papers on sleep staging from the same sensors. Related, not needed for AF.
- 12Papers that repeat #4 on new data with no new method.
Week 1
The signal
- Explain what a BCG beat is and why posture changes it
- Say how AF is diagnosed and what counts as an episode
1BCG tutorial2AF for engineers
1h 15m3 of 3 right
Week 2
Beats and intervals
- Find beats in a noisy bed signal
- Compute irregularity features over 30-second windows
3Bed-sensor intervals4Irregularity features
1h 30m1 of 3 done
Week 3
Data and evaluation
- Split data by subject, never by window
- Read a lab result and say what will change at home
5Open BCG dataset6Sleep-lab AF detection
1h 10mNot started
Week 4
The frontier
- Judge the current home benchmark and its claims
- Pick one open problem for your project
7120-night home study8Short-episode failures
1h 20mNot started
Reading time per weeksized to your 4 weeks
Week 1 · The signal75 min
Week 2 · Beats and intervals90 min
Week 3 · Data and evaluation70 min
Week 4 · The frontier80 min
Question 1 of 12Week 2 · from paper #4
Rate this question
Your progress12 questions
Week 1 · The signal3 / 3
Week 2 · Beats and intervals1 / 3
Week 3 · Data and evaluation0 / 3
Week 4 · The frontier0 / 3
- Every question is written from a takeaway extracted from one paper, and every answer links to the quote.
- No trivia: no years, no author names. Only why and how.
Reading list
It starts from the classics, not from last month.
- Classics pulled in through the citation backbone
- Every item typed (review, method, dataset) and rated for difficulty
- Every item readable: an open-access copy or a preprint, found for you
Classics found through citationsatlas papers citing each as background
All four predate the atlas window. None would appear in a keyword search over recent papers.
Curriculum
An order that respects what depends on what.
- Prerequisites from citation intent, not from dates
- A weekly plan sized to the time you have
- Learning goals for every week
Paper#4 Irregularity features for AF detection (2019)
Cites#2 Atrial fibrillation for engineers (2014) background
Quote"…we adopt the episode definition of [7]: at least 30 s of irregular rhythm…"
Read #2 firstBackground citations become prerequisites.
Paper#7 120-night home study (2026)
Cites#6 Sleep-lab AF detection (2024) compares
No order neededA comparison is not a prerequisite.
Self-test
A quiz that asks why, not what year.
- Why and how questions from extracted takeaways
- Every answer linked to its quote and section
- Testers rate each question, and the ratings feed the golden set
How should you split an overnight BCG dataset to estimate accuracy on new users?
ARandomly, by 30-second window
BBy subject, so no one is in both train and test
"…accuracy on window-level random splits was 9 points higher than on subject-level folds, which we recommend…"
#5 Open overnight BCG dataset · §4.1
Built on the atlas
A path through the field, not a pile of search results.
Citationswhat each paper builds on, so the order makes sense
Paper typereviews, methods and datasets, by difficulty
Open accessa free copy wherever one exists
Takeawaysthe key points of each paper, quoted
Who it's for
For anyone starting somewhere new.
New grad student
"My advisor said 'read the literature'. Which literature, in what order?"
Engineer changing fields
"I know signal processing. I don't know cardiology."
Manager getting up to speed
"I need to ask my team good questions by Friday."
More on the same atlas
Start any topic on the right foot.
In development, starting with sleep and cardiac sensing research. Tell us the project you are starting.