Proprietary instrument
How MDSoftune reads a thesis, word by word
MDSoftune is our own uniqueness engine, built with REDENN Informatics Inc., Canada. It reads a draft the way an examiner does — word by word, then phrase by phrase — and a report ships with every version, long before your university runs a check of its own.
- Built with REDENN Informatics Inc., Canada
- One report per version
- UGC 2018 · under 10%
- You remain the sole author
The problem
The problem
Why word level
Two ways a block-level check gets a medical thesis wrong
Ordinary similarity tools reach one verdict for a paragraph. A medical thesis is close to the worst possible document for that, because it fails in both directions at the same time.
Medical writing reuses standard language by necessity. Drug names, dosages, inclusion and exclusion criteria, scale definitions, the sentence that states how a sample size was calculated: these are written the same way by everyone because accuracy leaves no room to be original about them. A thesis that avoided that phrasing would be a worse thesis.
A block-level comparison cannot separate that necessary reuse from the thing you actually need to find. It has one verdict to give per block, and both of the following land in the same block.
Failure one · over-flagging
Routine phrasing is scored as if it were borrowed.
Take a sentence of the kind every method chapter needs: patients were randomly allocated into two groups of thirty each using a computer-generated table of random numbers. You wrote it. So did everyone who did the same thing, because there is only one accurate way to describe it. A block-level tool reaches its verdict on the paragraph that contains the sentence, so a paragraph that is overwhelmingly your own comes back looking reused, and you spend a week rewriting prose that was never the problem.
Failure two · under-flagging
A borrowed fragment hides inside a sentence you wrote.
Now take twelve words lifted from a review article and set in the middle of a forty-word sentence that is otherwise yours. Measured as part of a three-hundred-word block, those twelve words barely move the score. This is the passage that would embarrass you in front of an external examiner, and it is precisely the passage a block comparison is least able to see.
The two failures have opposite cures, which is why one coarse setting cannot fix both. Turn the sensitivity down and the hidden fragment survives. Turn it up and your own method chapter lights up like a fault. The only way out is to reach a verdict per word, and then ask which phrase each word belongs to — because a phrase boundary is what separates standard clinical language from a borrowed sequence.
The example sentences above are our own illustrative wording, written for this page. They are not quotations from anyone's thesis. See the difference drawn, in the next section.
MDSOFTUNE
BUILT WITH REDENN
INFORMATICS INC.
CANADA
§ II — MDSOFTUNE
Proprietary instrument
An examiner reads word by word. So do we.
MDSoftune is our own engine, developed with REDENN Informatics Inc., Canada. It reads a draft the way an examiner does: word by word, and phrase by phrase. Block-level tools compare paragraphs, so they miss a borrowed fragment sitting inside a sentence you wrote yourself, and they over-flag the routine clinical language that every thesis has to reuse.
A report ships with every draft, so you see what was flagged and why long before your university runs Turnitin or Drillbit. We work to the UGC 2018 convention of under 10% similarity.
WORD + PHRASE LEVEL
Not paragraph blocks.
EVERY DRAFT, EVERY VERSION
Re-run on each revision.
MEDICAL PHRASING AWARE
Standard clinical language is not treated as theft.
INCLUDED IN EVERY FEE
Never a paid extra.
What the engine actually does
- 01
Ingest
The draft is taken apart: sentences, phrases and single words, with tables, references and headings separated so they are judged by the right rule.
- 02
Scan
Every word and every phrase is compared, not paragraph blocks. A borrowed fragment sitting inside a sentence you wrote yourself is still found.
- 03
Classify
Routine clinical and statistical phrasing is marked as allowed. A sequence that sits too close to a published sentence is marked for rewriting, in your own voice, with the citation kept.
- 04
Report
A report ships with the draft: what was flagged, why, and what changed. Nothing is a surprise when your university runs Turnitin or Drillbit.
THE BOUNDARY
MDSoftune is used to prove that original work is original. It is never used to disguise copied content, to paraphrase around a source, or to defeat an academic-integrity check. A flagged passage is rewritten in your own voice with the citation kept in place, or it is cited properly and left alone. You remain the sole author.
MDSOFTUNE · UNIQUENESS REPORT
COMPLETE
SPECIMEN · OUR OWN ILLUSTRATIVE SAMPLE · MDT-26-0142
¶ 2.2 · REVIEW OF LITERATURE
Intrathecal clonidine has been reported to prolong both sensory and motor blockade in a dose-dependent fashion,1 with the most consistent benefit described between 30 and 75 micrograms.
¶ 3.1 · AIMS AND OBJECTIVES
Primary objective: to compare the duration of sensory blockade, defined as the time from intrathecal injection to two-segment regression2 of the sensory level.
¶ 4.3 · SAMPLE SIZE
Sample size was calculated to detect a 20-minute difference in two-segment regression time, assuming a standard deviation of 30 minutes, with 80% power and an alpha of 0.05.3
¶ 4.6 · STATISTICAL ANALYSIS
Continuous variables were compared using an unpaired t-test or the Mann-Whitney U test4 according to the distribution of the data.
- [1]
“in a dose-dependent fashion”
standard clinical phrasing — allowed - [2]
“the time from intrathecal injection to two-segment regression”
defined term — keep the wording, cite the source - [3]
“with 80% power and an alpha of 0.05”
standard statistical phrasing — allowed - [4]
“compared using an unpaired t-test or the Mann-Whitney U test”
close to a published sentence — rewrite
ILLUSTRATIVE SIMILARITY
1.8 per cent in this illustration
TARGET UNDER 10% · UGC 2018 CONVENTION
- WORDS CHECKED
- 98
- PHRASES FLAGGED
- 4
- TO REWRITE
- 1
ILLUSTRATION OF METHOD · OUR OWN SAMPLE TEXT · NOT A SCORE FOR ANY THESIS
BLOCK LEVEL · WHAT ORDINARY TOOLS SEE
One verdict per block. A borrowed fragment inside a paragraph you wrote yourself can pass, and routine clinical language can be condemned with it.
WORD AND PHRASE LEVEL · WHAT MDSOFTUNE SEES
One bar, one word, drawn at the real word lengths of the sentence above. The bracket is a phrase: the unit a block-level tool never looks at.
The report
The report
The deliverable
What the report contains, and when you get it
The scan is not the product. The report is, because it is the thing you and your writer actually act on — and the thing that stops a similarity figure from being a number nobody can explain.
What the MDSoftune report contains
One per version
- Similarity summary
- One figure for this version, with the previous version's figure beside it, so you read movement rather than a number in isolation.
- Flagged phrase list
- Every flagged phrase with its page and paragraph, and one of three verdicts: standard clinical phrasing, leave it alone; defined term, keep the wording and cite the source; distinctive sequence, rewrite it.
- A reason for each flag
- One line per flag, written to be read rather than scored, so you can tell whether the writer's judgement was right.
- What was set aside
- Your reference list, quoted and attributed material, tables and headings are listed separately, so you can see what was not counted towards the figure and why.
- Version, date and author
- The version number, the date the scan ran and who released it. The report is part of your record, not a loose file in a chat.
- What is still open
- Any flag the writer has not yet acted on, carried forward so it cannot quietly disappear between two drafts.
No report of ours states a conclusion about your integrity, and none of them awards a pass. A report lists what was found, where, and what to do about it.
When it reaches you
Three points
With every version
A report is attached to each draft in your dashboard at the moment that draft is released to you — not once, at the end, when it is too late to act on it.
Before you read the draft
The writer works the rewrite verdicts first, so the chapter you open has already been through the report rather than waiting for you to find the problems.
Once more before delivery
Plagiarism check is one of the fifteen named stages on your tracker. The final version is re-run at that stage, and that final report is the one you keep.
The check is a stage on your tracker rather than a favour you have to request, and it is included in the fee for every service. It is never sold as an extra.
Two checks
Two checks
Turnitin, iThenticate, Drillbit, Urkund
How this relates to the check your university runs
We are affiliated to none of them
Your institution runs its own similarity check at submission. That is the one that counts. MDSoftune does not replace it, cannot see it, and is not trying to beat it.
MDSoftune · our report
Before you submit
- Who runs it
- We do, on our own copy of your draft, before the draft reaches you.
- What it looks at
- Every word and every phrase of the draft, with quoted definitions, references, tables and headings separated out first so each is judged by the right rule.
- What it decides
- Which phrases to rewrite, which to keep and cite properly, and which are standard clinical language to leave alone.
- What it is not
- It is not your university's check. We have no access to any vendor's system or index, and no report of ours appears on your institution's record.
Your university · the check that counts
At submission
- Who runs it
- Your department, library or guide, at submission, on Turnitin, iThenticate, Drillbit or Urkund.
- What it looks at
- Your submitted file against that vendor's own index, under whichever settings your institution has chosen.
- What it decides
- The figure that goes on your record, and whether your submission is accepted as it stands.
- Why the numbers differ
- A different index, different settings and different exclusions produce a different figure. We do not predict that figure and we do not promise one.
What MDSoftune is actually for
The absence of surprises. By the time your department runs Turnitin or Drillbit, every version of your thesis has already been read word by word, the passages that would have been flagged have been rewritten in your own voice or cited properly, and you know which sentences are standard phrasing you are entitled to keep. You walk into that check knowing what is in your own document.
What it is not for is producing a number that matches theirs. Two tools with different indexes and different exclusion settings will disagree, and anyone who tells you otherwise is selling you something. We state our own figure, we say it is ours, and we make no promise about any score on any system we do not operate.
Vendor names appear here because they are the tools Indian health universities name in their own submission rules. MDThesis is not affiliated with, endorsed by, or acting for any of them.
Under 10%
Under 10%
UGC 2018
What under ten per cent actually means
It is the most quoted number in Indian postgraduate research and the least explained. Here is what the regulation says, and what it does not say.
The University Grants Commission's 2018 regulations on academic integrity establish the convention that similarity below ten per cent is treated as acceptable, and define graded penalties in bands above that level.[3] Most health universities apply the threshold to the postgraduate thesis and ask for a vendor report at submission. Where a university's own ordinance is stricter than the convention, its ordinance is the rule we work to.
- It is a convention about similarity, not a measure of honesty.
- A score of four per cent with one uncited borrowed paragraph is a far more serious problem than twelve per cent that is all correctly attributed method and statistics. The figure is a prompt to look, not a verdict on your integrity.
- What counts towards it depends on the settings.
- Whether the bibliography is excluded, whether quoted material is excluded, and the minimum match length below which a match is ignored are all choices your institution makes. The same thesis can return two different figures under two sets of settings. Ask your department which ones they run, and ask before you submit.
- Under ten per cent is a ceiling, not a target.
- Three kinds of text routinely make a sound thesis look worse than it is: the reference list, properly quoted and attributed definitions, and the standard phrasing of method and statistics. Removing a correct citation to bring a number down makes the thesis worse and the breach real.
So the working target we hold ourselves to is under ten per cent, checked on every version rather than discovered at the end. What we will not do is promise you a figure. The number depends on your institution's tool, its index and its settings, and a service that guarantees a percentage is guaranteeing something it does not control.
For DNB and DrNB candidates the integrity convention sits alongside the NBEMS timetable — protocol within 180 days of joining, thesis by 26 months, within the published page limit.[2] Both are loaded into your tracker at onboarding, because a thesis that is original and late still fails.
The platform
The platform
The rest of the machinery
Everything else, stated as a mechanism
Not a feature list. Each line below is something the software does, described closely enough that you could check whether it is true — which is the only kind of claim worth printing.
- Row-level security, per row
- Every read happens as you. The database decides which rows come back, not the application layer, so a record you are not party to answers not found rather than refusing a request it has already understood.
- Drafts are streamed, not sent
- A draft is served inline through an authenticated route as private, no-store, with noindex and nosniff set on the response. There is no shared link to leak and no attachment sitting in an inbox.
- Watermarked to the reader
- The watermark carries the name and email of whoever opened the page, and it is drawn for that reader at view time rather than baked once into a stored file.
- Download gated on settlement
- The download route compares what is settled against the total and refuses with a plain payment-required response, in one sentence, until the two are equal. Reading a draft never depends on it.
- Version history, nothing overwritten
- Each draft is a numbered version with its own page count, date and annotation thread. An earlier version is never replaced, so you can always see what changed and when.
- Pinned annotations, threaded
- A pin holds a page, a position on that page, a type — edit, add, remove, question or comment — the text it sits over, a status of open, in progress or resolved, and a reply thread. Your note and the reply to it stay attached to the spot they are about.
- Fifteen tracked stages
- The record moves through fifteen named stages from enquiry to closed, each with a plain sentence saying what is happening and who holds it. Nothing about your project lives only in someone's memory of a phone call.
- Milestones against statutory dates
- Every milestone carries a due date and a status, and the tracker counts the days remaining to each one. For DNB and DrNB trainees we set those dates from the NBEMS rule — protocol within 180 days of joining, thesis by 26 months — so a deadline is a dated row on your record rather than a reminder someone has to remember to send.
- Published fees, local currency
- Your country is read once from the edge request, mapped to one of seventeen published currencies and remembered in a cookie you can change at any time. The rupee figure is the published one; the conversion is arithmetic on top of it.
- Mumbai region
- Your record and your documents are held in the Mumbai region, under India's Digital Personal Data Protection Act 2023. The functions that read them run in the same region, which is a latency decision as much as a residency one.
The counts above — fifteen stages, seventeen currencies — are read from this platform's own configuration rather than stated as a claim.[6]
What we have deliberately not built
There is no paraphrasing tool. There is no button that rewrites a flagged passage for you, and there will not be one, because a sentence you did not write is not yours no matter which program produced it. There is no AI-detection score either: the available detectors are unreliable enough that publishing a number from one would be dishonest. Where a flag needs a rewrite, a person does the rewriting with you.
The same commitments, in summary
Protection of your work
Row-level security
Every table enforces row-level access. You read your own record, and nothing else.
View-only streaming
Drafts are streamed to you through an authenticated route, not handed over as a file.
Watermarked to you
Every page you read carries your own name and email across it.
Download gated
The final file unlocks when the fee is settled in full, and not before.
Mumbai region · DPDP 2023
Your record and your documents are held in the Mumbai region, so India's Digital Personal Data Protection Act 2023 applies to them.
Anonymised data only
We accept no patient identifiers. An NDA is available on request.
Instruments we work to
NMC PGMER-2023
The thesis obligations set out in the postgraduate medical education regulations.
NBEMS
DNB and DrNB protocol and thesis timelines, and the page limit, as published.
UGC 2018 · <10%
The academic integrity convention we work to on every draft.
ICMJE · Vancouver
Authorship criteria and reference style, applied as published.
No affiliation
We work to these published instruments. We are affiliated to none of the bodies that issue them.
Authorship and the uniqueness check
Sole author
Mentoring, editing, statistics and compliance. You remain the sole author of your thesis.
Not ghostwriting
We will not write your thesis for you, and we will not be named in it.
MDSoftune
Word-level uniqueness checking, built with REDENN Informatics Inc., Canada.
Every version
Each draft is checked word by word before your university sees it.
MDThesis is an independent academic mentorship practice. It is not affiliated with, endorsed by, or acting for the NMC, NBEMS, UGC or any university.
Free feasibility call
Tell us where your thesis stands. A senior doctor will tell you what to do next.
A senior doctor replies within one working day. No obligation. MD, MS, DNB, DrNB, DM, MCh, MDS and international programmes.
Request for a feasibility call
No obligation
Reply within one working day · No obligation · Your details are not shared
Prefer to write to us first? Contact the practice. We mentor and edit; you remain the sole author of your thesis.
Document: MDSoftune and the platform · Revision 2 · Last reviewed
Issued by MDThesis, a brand of REDENN Informatics Private Limited