This miracle doctor just wants to be fired!

The seventh patient's problem was even more pronounced.

The patient's coronary arteries were stable after the surgery, all examinations were good, and the model indicated a very low risk.

However, the patient lives in a mountainous area more than 100 kilometers away from the hospital, and the nearest county hospital is a two-hour drive away.

When he experiences chest pain after being discharged from the hospital, he usually tries to endure it at home first, and only asks his son to drive him to the hospital when the symptoms become unbearable.

Information such as transportation distance, accessibility to medical care, and whether the family owns a car is also not available in the database.

Zhao Sisi checked for several days in a row, and more and more blank spaces began to appear in the originally beautiful model framework.

……

Friday afternoon.

When Zhao Sisi went to find Zhou Cheng with her laptop, her expression was much more serious than when she first gave her report.

"Teacher, I've discovered that the model might be missing some variables."

Zhou Cheng was reviewing the complex calcification cases compiled by Li Mingyuan. Hearing this, he didn't look up: "What's missing?"

"The patient's living conditions, access to follow-up visits, family support, and medication habits are basically not included in the database."

Does it matter?

"It could be very important." Zhao Sisi pulled out a chair and sat down, explaining Mr. Wu's situation. "His traditional clinical risk isn't high, but he actually visits the doctor a lot. The model can only show the examination and surgical data; it can't see that he opens his shop every day at dawn, nor does it know that he often forgets to take his medication."

Zhou Cheng flipped to the patient's hospitalization record, looked at it for a while, and asked, "What are you going to do?"

"We increased the number of patient-reported outcomes and social factors surveys, but we couldn't make up for the lost data from previous retrospective studies. As a result, the advantage of the original sample size of over 10,000 cases was lost."

Zhao Sisi looked visibly frustrated as she spoke.

She spent a lot of time designing the model, thinking that the larger the database, the more reliable the result would be.

It wasn't until he actually spent a week following the patient that he realized the boundaries of the database were far more obvious than he had anticipated.

Zhou Cheng closed the medical record: "Content not in the database can still affect patients clinically. What the model can learn depends on what you give it. Missing variables don't disappear just because they're not recorded; they'll only turn into errors that you can't explain in the end."

Zhao Sisi nodded, then after a moment asked, "Should I still pursue this direction?"

"We can keep the direction in mind, but we need to narrow down the problem," Zhou Cheng said. "Let's not rush to predict the long-term outcomes for all patients. Let's see if we can start from a more specific and relatively complete data point, such as unplanned visits 30 days after surgery, or establish a prospective cohort to collect information that was missing in the past."

These few words did not provide a direct answer, but they gradually clarified Zhao Sisi's previously chaotic thoughts.

She turned on her computer and added a column next to the research framework, titled "External Variables in the Database".

……

On the same day, Li Mingyuan was also adjusting his animal experimentation protocol.

Most of the ten patients he tracked had severe coronary artery calcification.

As originally planned, he focused on recording the curvature of the calcification, its thickness, the length of the lesion, and the passage of instruments.

A few days later, he discovered that the factors that determine the treatment in clinical practice are far more complex than the model suggests.

Even with the same 360-degree circumferential calcification, some lesions are located in a straighter position, allowing the lithotripsy balloon to be delivered smoothly.

Some are located at obvious angles, making it difficult for even equipment to pass through.

The patient's kidney function, blood vessel diameter, previous stent placement, and financial capacity will also influence the final choice.

Complete animal experiments are certainly valuable, but before that, many issues in clinical cases have not yet been clarified.

He reassessed the three-year timeframe and ultimately decided to remove the twelve-month long-term pathological observation, splitting the project into two phases.

The first phase involves establishing a clinical database of complex calcified lesions, combined with the selection of intracavitary imaging analysis instruments and their immediate effects.

Animal experiments are only used for feasibility studies to verify several key mechanical issues.

As for complete long-term animal research, we will apply for a special project to continue it after the model matures.

……

More than half a month has passed.

When the three of them sat back down in the meeting room, the plans on the table were noticeably thinner than they had been a week ago.

Li Mingyuan deleted most of the animal experiment content.

Zhao Sisi's model no longer aims to predict all long-term risks at once.

Chen Yu changed the original "Study on Factors Affecting Medication Adherence" to "Study on Discharge Medication Information Comprehension Barriers and Intervention in Heart Failure Patients".

Zhao Sisi was taken aback when she saw Chen Yu's new title: "Are you preparing a medication prescription?"

"First, we need to assess whether the patient can understand the prescription," Chen Yu said. "The large number of medications is only one factor. Vision, living alone, duplicate medication names, different hospital prescription formats, and multiple adjustments to the treatment plan after discharge all increase the risk of errors."

She spread out several prescriptions for Mr. He's medications on the table.

After a few people looked at them together, they quickly discovered that one of the records used a generic name, the other used a brand name, and the community hospital's record used abbreviations.

If a certain medication is required to be discontinued on the first prescription but is reinstated on the third, it is difficult for someone without a complete medical history to quickly discern the difference.

After reading it, Li Mingyuan also changed his initial dismissive attitude: "This is indeed easy to eat incorrectly."

"Old He has already repeated taking the medication," Chen Yu said. "His medical records always say he has poor compliance, but he takes it every day, just not the right way."

After listening to the three people's reports, Zhou Cheng did not ask any further questions about where the papers could be published, nor did he offer his opinion on which direction was more advanced.

He placed the three revised proposals side by side on the table and looked at them for a few minutes.

"It's a little better now than last time."

Zhao Sisi asked, "Only a little?"

"At least we're starting to see patients."

Lin Feng, sitting next to him, couldn't help but chuckle.

None of the three students felt that the statement was sarcastic.

Over the past week, after actually accompanying patients through examinations, discharges, and follow-ups, they discovered that many of the problems written in the research protocol only existed in the researchers' imagination.

Other seemingly trivial difficulties affect patients' treatment outcomes every day.

Zhou Cheng returned the plan to them:

"Mingyuan first stratifies the cases, compressing the animal pre-experimentation to answer one key question."

"Sisi redefined the outcome indicators and clearly listed the information from outside the hospital that she planned to collect."

"Chen Yu, conduct small-sample interviews first, don't rush to design interventions. Revise the plans for all three people for another week."

The group meeting ended close to 7 p.m.

After Zhou Cheng finished signing the last medical record, there were still more than a dozen unread messages on his phone.

The three students each sent their revised frameworks. Lin Feng inquired about the plan for a left main coronary intervention the next day, and the School of Engineering sent another table of catheter stress model parameters.

He didn't reply immediately; instead, he shut down his computer, grabbed his coat, and left the hospital.

There's something I promised to do tonight.

The style of the wedding invitations had been chosen for a long time. Lin Wei made an appointment with the design shop in advance, asking him to come over after get off work to confirm the paper, font, and final copy.

This is the latest chapter.