The Father Who Defied Heaven
Chapter 95 — The Case That Still Looked Like Her
At 6:01 in the morning, Lu Xiaoxiao drew a person.
Large head.
Two arms.
Two legs.
One enormous bow.
Jian Wei looked over her shoulder.
“Who is that?”
“Nobody.”
He stared at the page.
Then at Xiaoxiao.
Then back at the page.
“Is that Madam Zhou?”
“No.”
The figure had Madam Zhou’s exact apron.
And the exact hairpin.
And the exact basket she used every morning.
Silence.
Madam Zhou entered.
Saw the drawing.
Stopped.
“That’s me.”
Xiaoxiao looked offended.
“No name.”
There.
Excellent.
Jian slowly pointed at the apron.
“Identity feature.”
Then the hairpin.
“Identity feature.”
Then the basket.
“Identity feature.”
Xiaoxiao frowned.
“But no name.”
Madam Zhou smiled.
“Still me.”
There.
Lu Chen drank tea.
Good.
Xiaoxiao took an eraser.
Removed the hairpin.
Jian looked again.
“Still Madam Zhou.”
She erased the apron pattern.
“Still Madam Zhou.”
She erased the basket.
He stared.
Then:
“Now maybe generic auntie.”
Madam Zhou nodded.
“Better.”
Good.
Xiaoxiao thought.
Then added a giant label:
NOT MADAM ZHOU
Silence.
Jian covered his face.
No.
At 6:17, Su sent her morning log.
No central symptoms.
No Gate cooling overnight.
Good.
One spontaneous warm pulse at 2:44 AM.
Peripheral.
Natural taper.
No abnormal cold.
Then:
No deliberate Blood Calling yesterday.
Good.
Next:
Question.
Lu Chen waited.
If you remove my name but keep Nine Cold Nails, Three Winter Gates, Mother, the vessel and Winter Remission, is that de-identified?
He replied:
No.
Three dots.
Then:
Good.
Another message:
How many rare details can stay before it becomes obviously me?
There.
Good question.
Depends on the audience and the combination.
A pause.
Annoyingly non-numeric.
Yes.
At 6:25, Meiyin sent:
Canglan’s next section calls this contextual recognizability.
Good.
Then:
He distinguishes direct identifiers from unique clinical signatures.
Excellent.
Another:
Tianhe wrote: “THE NAME MAY BE GONE WHILE THE PATIENT IS STILL STANDING IN THE ROOM.”
There.
Very good.
At 6:33, Zhao sent Qin Zhenguo’s baseline.
Stable.
Ruoxi stable.
Five meters.
Five minutes forty-five seconds repeatedly clean.
Six minutes remained the mild warmth threshold.
Today—
five minutes forty-five seconds.
Good.
No progression.
Approved.
Qin:
THE KINGDOM HAS BEEN DE-IDENTIFIED.
Ruoxi:
You are the only person here who calls it a kingdom.
Silence.
Exactly.
Good.
At 6:42, Patient Two sent:
OBSERVED
NO ROLE USE.
NO BRIDGE PRESSURE.
Then:
QUESTION.
IF SOMEONE REMOVES MY NAME BUT LEAVES A VERY RARE COMBINATION OF DETAILS THAT PEOPLE KNOW BELONG TO ME, IS THAT STILL PRIVATE?
Good.
Lu Chen replied:
Not necessarily. De-identification has to consider whether the remaining combination of details could still let the intended audience recognize you.
Patient Two:
SO PRIVACY DEPENDS PARTLY ON WHO IS LISTENING.
Yes.
Good.
At 6:51, Fan Qingshu sent:
Found Tianhe’s recognizability notes.
Then:
First margin: “ANONYMITY IS NOT ACHIEVED BY REMOVING THE LABEL FROM A UNIQUE BOX.”
Excellent.
Bring it later.
Fan:
WHITE CRANE?
Yes.
A pause.
I WILL ARRIVE ANONYMOUSLY.
No.
At 7:02, Xiaoxiao taped a new sign onto the wall.
NO NAME ≠ NOBODY KNOWS
Then underneath:
STORY CAN POINT
There.
Excellent.
Madam Zhou passed.
Read it.
Then pointed at the drawing.
“Exactly.”
Good.
Su arrived at 7:16.
One coffee.
Normal breakfast.
Small bag.
She saw the sign.
Then the drawing.
Then Madam Zhou.
Then Xiaoxiao.
“I assume the experiment already happened.”
“No.”
Maybe.
Baseline first.
Normal temperature.
Normal blood pressure.
No dehydration.
No unusual fatigue.
No chest symptoms.
No palpitations.
No dizziness.
No active cold.
Life Meridian stable.
First Gate—
left.
Quiet.
Second—
quiet.
Third—
quiet.
Winter remission stable.
Buffers One through Five fully supported.
Buffer Six full coupling.
Good.
No residual Gate cooling at rest.
Good.
Su watched.
“Eligible to be anonymized badly?”
“For travel.”
She laughed.
“Consistent.”
Deng arrived.
Same vehicle.
Same privacy.
Good.
Before leaving, Xiaoxiao handed Su the drawing.
Now mostly generic.
At the bottom she had written:
MAYBE NOT MADAM ZHOU
Good.
The drive to White Crane remained uneventful.
No attack.
No unknown compatible source.
No rightward First Gate diversion.
No paternal contingency trigger.
No one recognizing Su from a conference abstract that did not yet exist.
Good.
At 8:40, Su said:
“I think recognizability has at least three layers.”
Good.
“What?”
“Direct identifiers.”
“Yes.”
“Rare details.”
“Yes.”
“And combinations of individually harmless details.”
There.
Excellent.
“Four.”
She looked at him.
“What?”
“The audience’s prior knowledge.”
She nodded.
“Yes.”
Good.
“Five.”
She sighed.
“What now?”
“Timing.”
There.
Interesting.
A case described immediately after a known unusual event may be recognizable even if the same details years later would be less obvious.
Good.
Su stared out the window.
“So context itself identifies.”
“Yes.”
There.
At 9:25, White Crane Temple.
Ordinary morning.
A novice carrying folded paper.
Someone trimming medicinal vines.
Thin clouds moving above the roofline.
The old dog lying beside the threshold and forcing everyone to step around him.
Good.
Meiyin waited in the records room.
No restricted files on the table.
Good.
Only Ye Canglan’s document.
Rest first.
Twenty minutes.
Water.
No privacy exercise yet.
Good.
At 9:58, repeat baseline.
Stable.
No active Call.
No shared field.
No preparation.
Good.
Meiyin placed the page down.
RECOGNIZABILITY AFTER REMOVAL OF DIRECT IDENTIFIERS
Underneath, Tianhe had written:
THE CASE THAT STILL LOOKED LIKE HER
There.
Good.
Definitions first.
Of course.
Direct identifier
Information that explicitly names or directly identifies the patient, such as name, exact personal identifier, address, or similarly direct identifying information.
Good.
Contextual identifier
A detail that may not identify the patient by itself but can become identifying when combined with the audience’s prior knowledge or other case features.
This content has been unlawfully taken from NovelPhoenix; report any instances of this story if found elsewhere.
There.
Rare-feature signature
A distinctive clinical, family, procedural, temporal, or physiologic pattern whose unusual combination may make the patient recognizable even without direct identifiers.
Excellent.
Audience-dependent recognizability
The likelihood that a particular audience could identify the patient from the remaining information based on what that audience already knows.
There.
Very important.
Temporal recognizability
Increased identification risk because the case is described close in time to a known event, treatment, admission, incident, or other publicly or professionally known occurrence.
Good.
Feature necessity
Whether a specific detail is actually required to preserve the teaching, analytic, or operational purpose of the case.
Excellent.
Generalization
Replacing an unnecessary precise detail with a broader category while preserving the relevant lesson.
There.
Suppression
Removing a detail entirely because it is not required and contributes unnecessary identification risk.
Good.
Residual recognizability
The remaining risk that the patient can still be identified after direct identifiers and unnecessary details have been removed or generalized.
Excellent.
Su read.
Then:
“So de-identification is not yes or no.”
“Correct,” Lu Chen said.
“It is a risk assessment.”
“Yes.”
Good.
Ye Canglan wrote:
Do not ask whether the name is gone. Ask whether the person is still visible.
There.
Excellent.
Tianhe margin:
A rare case can introduce itself.
Silence.
Strong.
Another:
The audience is part of the privacy calculation.
There.
Very good.
First exercise.
Take the proposed teaching description from Chapter 94.
Original draft:
Adult hereditary-resonance carrier with nine stabilizing implants, three directional protective gates, prior childhood resonance-vessel exposure, historical maternal proximity-dependent entrainment, current Winter Remission, and recurrent task-associated First Gate cooling.
Silence.
Obviously Su.
Good.
Meiyin asked:
“Direct identifiers?”
Su answered:
“None.”
Correct.
“Recognizable to ordinary physician?”
“Probably not.”
Reasonable.
“To White Crane senior staff?”
Su looked at the line.
“Yes.”
Absolutely.
“To Jian?”
Maybe not.
He knows some, but not all White Crane details? Jian did not go to White Crane; likely knows broad current issues but not restricted history. Still maybe could infer Su due nine nails and gates. But no need test on him. Good.
“To someone who treated me as a child?”
“Almost certainly.”
Meiyin said.
Good.
So audience matters.
Excellent.
Tianhe:
Anonymity is not measured against a hypothetical stranger if the actual audience is a room full of people who know the case.
There.
Excellent.
Second exercise.
Feature necessity.
Teaching objective:
A historical restriction should not automatically be reactivated when one current sign superficially resembles a prior syndrome. Compare defining features before assuming recurrence.
Good.
What details are required?
Need:
historical restriction.
current partial resemblance.
mechanistically important differences.
decision not to reactivate automatically.
Good.
Do we need:
Nine Cold Nails?
No.
Suppress.
Three Winter Gates?
No.
Suppress.
Blood Calling?
No.
Suppress.
Meiyin as mother?
No.
Suppress.
Vessel?
No.
Suppress.
White Crane?
No.
Suppress.
Winter Remission?
Maybe not.
Could generalize:
current stabilized state.
Good.
Current First Gate cooling?
Could generalize:
mild current physiologic warning sign.
Good.
Historical proximity-dependent entrainment?
Could generalize:
previous condition-specific environmental trigger.
Good.
There.
Teaching point survives.
Excellent.
New vignette:
A patient with a previously documented condition-specific environmental restriction later develops a mild physiologic sign that superficially resembles one feature of the old syndrome. Review shows the current event lacks several defining features of the historical mechanism. The prior restriction is therefore not automatically reinstated; current monitoring continues while the new sign is characterized separately.
There.
Good.
Su read.
“Could anyone identify me from that?”
Maybe someone very suspicious, but not likely based solely on this abstracted form.
Good.
Meiyin said:
“I could not reliably identify you from that if I had not participated in writing it.”
There.
Useful.
Su nodded.
“Better.”
Good.
Tianhe margin:
Generalize the detail, not the lesson.
There.
Excellent.
Another:
If precision is not doing educational work, precision is probably doing privacy harm.
Silence.
Strong.
At 10:31, Qin and Ruoxi began five minutes forty-five seconds at five meters.
Minute one.
Stable.
Two.
Awareness.
Three.
Mild.
Four.
Stable.
Five.
No warmth.
Five-thirty.
No warmth.
Five-forty-five.
No warmth.
End.
Clean.
Good.
No progression.
Third exercise.
A detail that is necessary.
Suppose teaching objective changes.
Now the specialist audience wants to understand:
How can a protective gate remain intact and selectively responsive during remission rather than simply becoming inactive?
Interesting.
That teaching point may actually require some gate-specific physiology.
Good.
Can we generalize away all distinctive details?
No.
Need at least:
protective gate.
remission.
selective response.
native versus harmful trigger discrimination.
Good.
But do we need:
three gates?
Maybe not.
Can say:
a prior protective meridian mechanism.
Good.
Need Nine Nails?
No.
Need Su’s exact disease?
Maybe can say:
suppression syndrome in remission.
Good.
Need Blood Calling?
Maybe not.
Could say:
hereditary resonance disorder.
Still somewhat specialized but less unique.
Good.
Need maternal proximity?
No.
Need vessel?
No.
Excellent.
Draft:
In a patient with a hereditary resonance-related suppression syndrome in remission, a previously protective meridian response remained structurally intact and could still activate selectively under relevant stress without remaining continuously engaged. This illustrates that remission can coexist with preserved protective architecture.
There.
Good.
Specialist lesson preserved.
Identifying signature reduced.
Excellent.
Su read.
“That feels less like me.”
“Yes.”
Good.
But would White Crane seniors still possibly recognize?
Maybe.
They might know few such cases.
Need audience-dependent residual risk.
Good.
Meiyin said:
“The older specialists could still guess.”
There.
Important.
So what next?
Could restrict audience further?
Actually smaller specialist audience may increase recognizability, not reduce. A broader outside audience might know less. Interesting.
Good.
Su frowned.
“The people who most understand the mechanism are also the people most likely to recognize the case.”
Exactly.
There.
Hard problem.
Tianhe anticipated.
Expertise can increase re-identification risk because experts know which rare details matter.
Silence.
Excellent.
Another:
The most technically appropriate audience may also be the most contextually identifying audience.
There.
Strong.
So sometimes you cannot solve with generalization alone.
Need perhaps patient permission.
Or no use.
Good.
Fourth exercise.
Could you restrict the audience to fewer people?
Counterintuitive.
If those few already know Su, recognizability may be near certain.
So audience restriction reduces spread but not identification.
Good distinction.
Su said:
“So smaller audience does not automatically mean anonymous.”
“Correct.”
Good.
It may still be privacy-protective because fewer recipients.
But not de-identification.
Excellent.
Ye Canglan:
Confidentiality and anonymity are different protections.
There.
Important.
Confidential use
Identified or recognizable information is used only within a restricted authorized audience.
Anonymous or sufficiently de-identified use
The audience cannot reasonably identify the patient from the material.
Good.
Su read.
“So if White Crane specialists would recognize me, the case can still potentially be used confidentially with my permission.”
“Yes.”
Good.
“But we should not call it anonymous.”
“Correct.”
There.
Excellent.
Tianhe margin:
Do not call restricted disclosure anonymous merely because the room is small.
There.
Very good.
Fifth exercise.
False de-identification.
Fun.
Fan arrives early? Not yet. Could use Meiyin's examples.
Example A:
Patient A
instead of Su’s name.
Then full details.
Not de-identified.
Good.
Example B:
change age from exact to broad adult.
Still full rare signature.
Not enough.
Example C:
remove dates but say:
the only surviving patient from the fourth post-gate vessel cohort with nine Cold Nails and Three Winter Gates.
Obviously identifiable.
Bad.
Su stared.
“That one would be worse than my name.”
Maybe.
Good.
Example D:
replace Meiyin with:
a first-degree maternal relative
Still perhaps recognizable in context.
Yes.
Example E:
say:
a compatible family member
Better.
But combined with nine nails, three gates, vessel history, still recognizable.
Good.
So no single-field approach.
Need combination.
Excellent.
Tianhe:
De-identification is not a game of changing one noun at a time until the author feels less guilty.
Silence.
Very Tianhe.
Good.
Su laughed once.
“That one was personal.”
Probably.
At 11:07, Patient Two sent:
QUESTION.
IF SOMEONE CHANGES MY NAME TO “PATIENT A” BUT LEAVES MY VERY UNIQUE HISTORY, IS THAT ACTUALLY DE-IDENTIFIED?
No.
Lu Chen replied:
Not necessarily. A label like “Patient A” does not protect privacy if the remaining details still make the person recognizable to the intended audience.
Patient Two:
SO CODE NAME IS NOT ANONYMITY.
Correct.
Good.
Lunch.
Ordinary.
Su sat across from Meiyin.
Then:
“Would you recognize me from ‘three protective gates in remission’?”
Meiyin thought.
“Yes.”
Good.
“What if it was ‘protective meridian architecture remained selectively responsive during remission’?”
Meiyin thought longer.
“Possibly.”
Good.
“What if they removed hereditary resonance?”
“Less likely.”
Good.
“What if they removed remission?”
“Then the lesson changes.”
There.
Exactly.
That is the tradeoff.
Good.
Su looked down at tea.
“So sometimes the feature needed for the lesson is the feature that identifies.”
“Yes.”
Meiyin said.
There.
No easy solution.
Good.
Tianhe’s note:
When the educational value and the identifying value reside in the same detail, redaction may not solve the problem.
Silence.
Excellent.
Then what?
Options:
patient agrees to confidential identifiable/re-identifiable use.
Change teaching objective.
Use a more abstract principle.
Use a synthetic/composite example if possible and honest.
Or don't use.
Good.
Need define these.
Purpose redesign
Change the teaching question so fewer distinctive details are necessary.
Good.
Confidential identifiable use
Use recognizable details within an explicitly limited audience under appropriate permission and confidentiality.
Good.
Composite or synthetic vignette
Construct a clearly labeled educational scenario from generalized principles or multiple non-identifying patterns without falsely presenting it as one real patient's history.
Important.
Non-use
Do not use the case when meaningful de-identification is not possible and no sufficient basis exists for identifiable use.
Excellent.
Su read the third.
“Composite worries me.”
Good.
“Why?”
“It could create a fake patient that looks like one mechanism.”
Exactly.
Need guard against merging distinct mechanisms.
Great.
Lu Chen said:
“Do not combine unrelated mechanisms into a false clinical story.”
There.
Excellent.
If composite used, it must preserve only the principle being taught and clearly state it is constructed.
Good.
No blending Blood Calling with Patient Two or Qin etc.
Excellent.
Meiyin nodded.
“Otherwise teaching becomes fiction presented as evidence.”
There.
Strong.
Tianhe margin:
A composite can protect privacy. It can also manufacture a case that never existed. Label it honestly.
Excellent.
Another:
Do not solve confidentiality by falsifying mechanism.
There.
Perfect.
No composite needed today.
The abstract principle already works.
Good.
At 11:41, Qin and Ruoxi remained stable after their clean exposure.
No issue.
Good.
Sixth exercise.
Temporal recognizability.
Good.
Suppose the teaching is presented tomorrow at White Crane.
Everyone there knows yesterday there was an unusual restricted-file review.
Even a generalized case might be recognized because of timing.
There.
Good.
Su frowned.
“So waiting can reduce identification risk.”
“Sometimes.”
Good.
But not always.
Rare signature may persist for years.
Still.
If teaching can wait—
delay may help.
Good.
Tianhe:
Timing can identify even when content does not.
There.
Excellent.
Another:
A case presented the morning after an unusual event may announce its source through the calendar.
Good.
So the abstract teaching principle should not be introduced as:
Yesterday we had a case where…
No.
Just teach principle later without linkage.
Good.
Su nodded.
Excellent.
Seventh exercise.
Audience prior knowledge.
They create three hypothetical audiences.
Audience A — General clinicians outside White Crane
Low knowledge of Su’s case.
Could likely use moderately specific generalization safely.
Audience B — White Crane trainees who know the old family history
Higher recognition risk.
Need stronger abstraction or permission.
Audience C — The two clinicians who directly treated Su
Recognition inevitable.
Calling material anonymous would be meaningless.
If using within direct case review for current care, that's purpose continuity.
If teaching, it becomes identifiable use.
Need separate permission.
Good.
Su read.
“This is much clearer.”
Yes.
De-identification is not property of document alone.
It is document + audience + context + timing.
There.
Excellent.
Tianhe margin:
The same sentence can be anonymous in one room and identifying in another.
Silence.
Very good.
At 12:03, pre-travel baseline.
Stable.
No Gate cooling.
Good.
Back at clinic—
Xiaoxiao had redrawn Madam Zhou.
This time:
generic hair.
generic clothes.
No basket.
No apron.
No hairpin.
Madam Zhou looked.
“Who is that?”
Xiaoxiao smiled.
“Nobody.”
Good.
Jian squinted.
“Could be several aunties.”
Success.
Then Xiaoxiao added:
MAKES BEST PORK ADOBO ON THIS STREET
Madam Zhou immediately pointed.
“Me.”
Silence.
Back to work.
Good.
At 2:16, Fan arrived.
Clinic.
Good.
He brought Tianhe’s recognizability notes.
Saw Xiaoxiao’s drawing.
Read the label.
Then:
“Contextual identifier.”
Yes.
He sat down.
“Can I use my archive training example from Chapter 94?”
Su looked at him.
“What details are in it?”
“An archivist manages sealed-record identifiers, version control and access logs without reading clinical content.”
Good.
“Any location?”
“No.”
“Any specific patient?”
“No.”
“Any unusual mechanism?”
“No.”
“Any exact access event?”
“No.”
Good.
Low recognizability.
Fine.
Fan nodded.
“Excellent.”
Then:
“Can I say it was inspired by a real case?”
Su thought.
“Why?”
Silence.
Fan looked offended.
But fair.
“Teaching value?”
“None.”
“Then no.”
There.
Good.
Fan nodded slowly.
“That question is becoming inconveniently effective.”
Excellent.
Tianhe note:
If the detail “this really happened” adds drama but not instruction, consider removing it.
There.
Strong.
Fan sighed.
Fine.
Eighth exercise.
Changing one rare feature deliberately.
Could we alter facts to protect identity?
Careful.
If teaching case is presented as real, altering clinical facts risks falsification.
Good.
Su asked:
“Could they say eight nails instead of nine?”
“No,” Lu Chen said.
Immediate.
Good.
“Could say two gates?”
“No.”
“Different relative?”
“No.”
Good.
Because that would create false data if presented as actual case.
Excellent.
Could generalize:
multiple stabilizing implants
Yes.
Could generalize:
multiple protective gates
Maybe.
If exact number not needed.
Good.
Difference between generalization and fabrication.
Excellent.
Definitions:
Generalization
Truthful reduction in precision.
Example:
nine nails → multiple stabilizing implants.
Fabrication
Replacing a true specific fact with a false specific fact.
Example:
nine nails → eight nails.
Good.
Su nodded.
“That distinction matters.”
Yes.
Tianhe:
Privacy permits less precision. It does not require false precision.
There.
Excellent.
Another:
Broaden. Do not counterfeit.
Silence.
Good.
At 2:47, Patient Two sent:
QUESTION.
CAN SOMEONE CHANGE A DETAIL ABOUT ME TO A DIFFERENT FALSE DETAIL TO HIDE WHO I AM?
If presented as real clinical facts, not ideal. Need concise.
Lu Chen replied:
If the material is presented as a real case, replacing true specific facts with false specific facts can distort the record and teaching. It is usually better to generalize, suppress, or clearly label a constructed vignette rather than quietly inventing different clinical facts.
Patient Two:
SO “LESS SPECIFIC” IS DIFFERENT FROM “FALSE.”
Correct.
Good.
Ninth exercise.
Could combine cases?
Again.
White Crane has other hereditary-resonance patients historically, perhaps not in current story? We shouldn't invent details. We can discuss conceptually only.
Good.
Meiyin said:
“Canglan sometimes taught from constructed examples.”
Interesting.
Need not imply specific cases.
“He labeled them?”
“Yes.”
Good.
Example heading:
CONSTRUCTED TEACHING VIGNETTE — NOT A SINGLE PATIENT HISTORY
Excellent.
Then use broad physiological principles.
No hidden real patient.
Good.
Su nodded.
“That I can accept.”
Good.
But not needed today.
Fine.
At 3:09, He An sent:
MINE NOW:
IF I SAY “A FRIEND” BUT EVERYONE KNOWS I ONLY HAVE ONE FRIEND WITH A BLUE MOTORCYCLE, THEY KNOW WHO I MEAN.
Excellent.
Lu Chen replied:
Correct.
He An:
CONTEXTUAL IDENTIFIER.
Yes.
Then:
I HAVE MORE THAN ONE FRIEND.
Good.
No issue.
At 3:31, Luo Jian sent:
TODAY
I was told removal of direct identifiers does not guarantee de-identification.
Good.
Then:
I did not witness the teaching-case review.
Good.
Then:
I will consider the intended audience’s prior knowledge when assessing recognizability.
Excellent.
I will distinguish truthful generalization from quiet fabrication.
Very good.
I will not call a small confidential audience anonymous merely because disclosure is limited.
Excellent.
I will not assume “Patient A” is de-identified if the case details remain uniquely recognizable.
Very good.
I will prefer an abstract teaching principle when the lesson survives without the patient-specific story.
Excellent.
Lu Chen replied:
Continue.
At 3:54, clinic quieted.
Su stayed.
Of course.
She looked at the teaching drafts.
Original:
obviously her.
Second:
less obviously her.
Third:
abstract enough that the lesson remained and she largely disappeared.
Good.
She said:
“I thought de-identification meant deleting.”
“Yes?”
“It is more like asking which details are doing actual work.”
There.
Excellent.
Each detail either:
supports the lesson,
supports safety,
supports analytic validity,
or does not.
Good.
And if it does not—
remove or generalize.
Excellent.
But if it does—
then evaluate recognizability.
Good.
If recognizability remains high:
ask permission for confidential use.
change audience.
change purpose.
construct a clearly labeled general example.
or do not use.
There.
Good.
No magical anonymization.
Excellent.
Fan found another Tianhe page.
Of course.
The purpose of de-identification is not to make the document look anonymous to the author.
Silence.
Good.
It is to reduce the chance that the audience can identify the patient.
There.
Excellent.
Then:
Authors are poor judges when they already know who the patient is. Test against the audience.
Very good.
Su looked at that.
“How?”
Good question.
Could have independent privacy reviewer who knows audience context but not necessarily patient identity, assess recognizability. But don't need bureaucracy. Maybe simple checklist.
Good.
Ye Canglan had one.
RECOGNIZABILITY TEST
What is the smallest audience that will receive this?
What do they already know?
Which remaining details are rare?
Which combinations become distinctive?
Does timing make the source obvious?
Could someone familiar with the patient reasonably infer identity?
Which details can be generalized or removed without harming the purpose?
If recognizability remains high, is identifiable/confidential use actually justified and permitted?
If not, should the case not be used?
There.
Excellent.
Su read.
“Very clean.”
Yes.
At 4:17, Su experienced a spontaneous home pulse.
Warm.
Peripheral.
Natural taper.
No Gate cooling.
Good.
No one wrote:
rare hereditary carrier with Nine Cold Nails demonstrates spontaneous home pulse.
Excellent.
At 4:31, Meiyin called.
Video.
She had reviewed the abstract teaching draft.
“I would not recognize you from this without already knowing we had written it.”
Good.
Su said:
“Would you recognize the specialist version?”
“Yes.”
Good.
“Then that one is not anonymous.”
“No.”
Excellent.
Simple.
Meiyin added:
“The abstract version teaches the decision principle better anyway.”
Interesting.
Often true.
Good.
Su smiled faintly.
“Less story. More lesson.”
Yes.
At 4:49, Fan sent:
I HAVE AN IDEA.
Danger.
Then:
A review panel could include someone who knows the audience but not the patient.
Interesting.
Useful.
No need create panel routinely.
But for high-risk rare cases, independent recognizability review could help.
Good.
Lu Chen replied:
Sometimes.
Fan:
THIS IS PRAISE.
No.
At 5:04, Patient Two sent:
I THINK I UNDERSTAND SOMETHING.
Then:
THE QUESTION IS NOT “DID YOU REMOVE MY NAME?”
Good.
IT IS “CAN THE PEOPLE WHO WILL SEE THIS STILL FIGURE OUT IT IS ME?”
Excellent.
Lu Chen replied:
Correct.
At 5:19, clinic closed.
Su stayed another minute.
Then stood.
“Next chapter?”
Not fully set.
Need determine from emerging issue.
Could be audience that already knows too much.
Maybe focus on small audiences/confidential use vs anonymity.
Tianhe note perhaps.
Fan handed over the next Ye page.
Title:
DISCLOSURE TO AUDIENCES WITH PRIOR CASE KNOWLEDGE
There.
Good.
Tianhe’s handwritten title beneath:
THE ROOM THAT ALREADY KNEW HER
Silence.
Excellent.
Su read.
“That is next.”
Yes.
First line:
A case cannot be made anonymous to people who already know which patient produced the case merely by removing the name from the slide.
There.
Strong.
Second:
When the audience already knows, stop pretending anonymity is the protection. Use confidentiality, purpose limits, and permission instead.
Excellent.
Third:
Do not call recognition failure when recognition was inevitable. Call the use identifiable and govern it honestly.
There.
Very good.
That would be Chapter 96.
Good.
At 5:37, Qin and Ruoxi remained stable after their clean exposure.
No issue.
Good.
At 6:03, Xiaoxiao returned to her Madam Zhou drawing.
She removed:
the basket.
the apron.
the hairpin.
the adobo label.
Good.
Then drew three random people beside her.
Now four generic aunties.
Madam Zhou looked.
“Which one is me?”
Xiaoxiao smiled.
“Maybe none.”
Good.
Jian pointed.
“Now sufficiently generalized.”
No.
Then Xiaoxiao wrote:
ONE OF THEM MAKES VERY GOOD ADOBO
Madam Zhou narrowed her eyes.
“Still me.”
Silence.
Almost.
At 6:41, Patient Two sent:
FINAL ENTRY
NO ROLE USE.
NO BRIDGE PRESSURE.
Then:
SOMEONE WANTED TO REMOVE MY NAME AND USE MY CASE.
Good.
WE CHECKED WHETHER THE STORY WOULD STILL IDENTIFY ME.
Excellent.
IT WOULD.
Good.
SO THEY USED THE LESSON WITHOUT USING MY STORY.
Excellent.
At 7:18, Su messaged:
One last thought.
He waited.
Maybe a case stops being anonymous before anyone says the name.
Yes.
Another:
Sometimes the audience already knows enough to finish the sentence.
There.
Excellent.
Another:
Then pretending the name is the only secret feels dishonest.
Yes.
Lu Chen replied:
Correct.
Three dots.
Then:
Good.
A pause.
Of course.
Good.
Another message:
Good night, Doctor Lu.
Good night.
At 8:58, Xiaoxiao was in bed.
Rabbit under one arm.
She asked:
“If no name but everyone knows?”
“Then it is not meaningfully anonymous to that audience.”
She frowned.
Daddy words.
Again.
He simplified.
“If they can still tell it is you, hiding your name did not hide you.”
Her eyes widened.
“Oh.”
“Yes.”
“So change story?”
“Remove or generalize details that are not needed.”
“If lesson breaks?”
“Then maybe ask permission for identifiable confidential use.”
She frowned harder.
“Or?”
“Do not use the case.”
There.
She nodded.
“Can use lesson only.”
“Yes.”
Good.
She hugged the rabbit.
“Lesson can go without person.”
Often.
“Yes.”
He turned off the light.
Downstairs, Lu Chen opened Su Qingyue’s chart.
No restricted access today.
Good.
The paternal contingency record remained sealed.
The historical proximity source remained closed.
Excellent.
But the privacy problem had changed again.
At first, privacy meant controlling who received information.
Then controlling why they used it.
Now—
even when names were removed—
the information itself could still point back to Su.
There.
Nine Cold Nails.
Three Winter Gates.
Blood Calling.
Childhood vessel exposure.
Meiyin’s compatible proximity effect.
Winter Remission.
Task-associated Gate cooling.
That combination was not merely medical detail.
Inside the wrong room—
it was an identifier.
Good.
And the “wrong room” did not necessarily mean an untrustworthy room.
White Crane specialists could be excellent clinicians.
Respectful.
Professional.
Necessary.
But because they knew the history—
they were precisely the people most capable of recognizing the case.
There.
Trustworthiness did not equal anonymity.
Important.
So the correct protection changed depending on audience.
For strangers—
de-identification might be meaningful.
For experts with prior knowledge—
confidentiality and explicit purpose limits might be more honest.
For people who already knew Su personally—
pretending the case was anonymous would be absurd.
Good.
Lu Chen opened Tianhe’s final page.
A patient can remain visible after every name is erased.
Silence.
Good.
Then:
The shape of the story may be enough.
There.
And finally:
When the audience already knows the shape, stop promising invisibility. Promise boundaries instead.
Silence.
Enough.
That would be next.
Because de-identification had limits.
A room full of strangers might not recognize Su.
A room full of White Crane specialists might.
The exact same slide could be anonymous in one place and identifiable in another.
So the next question was not simply:
How do we hide her better?
It was:
What protection is honest when hiding her is no longer realistic?
Confidentiality.
Purpose limits.
Restricted attendance.
No recording.
No redistribution.
Explicit permission.
Perhaps no case use at all.
Different tools.
Different audience.
Different question.
Again.
If you find any errors (non-standard content, ads redirect, broken links, etc..), Please let us know so we can fix it as soon as possible.
ReportUse arrow keys (or A / D) to PREV/NEXT chapter
Loading comments…