LIS system checks how process functions fit with business needs

The difference in the needs of clinical and scientific research scenarios sets a clear Chuhe Han boundary for the LIS function design: the former needs to ensure zero errors and fast turnover under the high pressure of thousands of specimens per day, while the latter needs to achieve traceability and high adaptability in the ever-changing experimental design. This paper will dismantle the design logic, technical details and optimization paths of LIS functions in different scenarios along the whole process of inspection application, specimen acceptance, inspection execution, result review, and report release, combined with the practical experience of 23 tertiary hospitals and 11 scientific research institutions.

When outpatient nurses scan the code to receive routine blood specimens, when researchers retrieve tumor gene sequencing data from three years ago, and when clinicians view the patient’s examination trend chart for nearly half a year with one click in the HIS system, the Laboratory Information System (LIS) has long jumped out of the traditional framework of data entry tools and played the role of an invisible center to connect every node of the whole testing process.

1. Acceptance of inspection applications

The inspection application is the first button in the inspection process, and if it is buckled in the wrong position, the whole subsequent process will be affected. The standardization of clinical scenarios and the personalization of scientific research scenarios determine that this link must take the road of sub-track design.

1. Clinical Laboratories

The number of applications for clinical testing is often dozens of times that of scientific research – the laboratory department of a provincial tertiary hospital receives more than 5,000 outpatient + inpatient applications per day. Therefore, the real-time integration of LIS with HIS is not a plus, but a must.

1) Interface protocol

HL7FHIR protocol is the current mainstream choice, but to achieve synchronization within 3 seconds, it needs to be deeply optimized in three aspects:

1. Field mapping refinement: Synchronize not only the patient’s basic information (name, medical card number), but also the clinical context (such as outpatient/inpatient identification, department, diagnosis code). For example, after a hospital synchronizes the fever clinic logo to LIS, the system will automatically prioritize the marking of such specimens;
2. Message queue buffering: Using middleware such as RabbitMQ, when the peak of HIS concurrent requests (such as the outpatient peak at 10 a.m.) exceeds the processing capacity of LIS, the queue will be sorted by timestamp + urgency to avoid data loss.
3. Breakpoint continuation mechanism: If the network is interrupted, the system records the synchronized field positions and resumes transmission from the breakpoint after the connection is restored, instead of sending the full amount of data repeatedly.

The practice of the First Affiliated Hospital of a university shows that this architecture has stabilized the data synchronization success rate of HIS-LIS at 99.98%, and there is no application loss even within 30 minutes of the hospital network cutover.

What does a product manager need to do?
In the process of a product from scratch, it is not easy to do a good job in the role of product manager, in addition to the well-known writing requirements, writing requirements, writing requirements, there are many things to do. The product manager is not what you think, but will only ask you for trouble, make a request:

View details >

2) Intelligent pre-examination

The checks built into the system are not dead code, but a living dictionary maintained by multiple departments. A hospital’s rules engine contains three-level checks:

1. Compliance layerFor example, if a province stipulates that CTDNA testing is limited to patients with advanced cancer, the system will automatically compare the patient’s ICD-10 diagnosis code (such as 900 for advanced lung cancer), and if it is not the corresponding code, a medical insurance policy restriction warning will pop up and display the policy document number;
2. Rational layer: Combined with the clinical pathway dynamic adjustment, for example, the glycated hemoglobin detection interval in diabetic patients is 3 months by default, but if the patient has recently adjusted the insulin dose, the doctor can manually check the clinical necessity, and the system will record the reason and archive it;
3. Associations layer: Construct a project dependency network, such as when applying for four coagulation functions, the system will check whether to apply for blood routine at the same time (due to the number of platelets affecting the interpretation of coagulation results), if not applying, it will display the recommended supplementation, reason: abnormal platelet count may lead to misjudgment of coagulation results.

These rules are updated through a visual configuration interface (similar to Excel formula editing), and laboratory directors, medical insurance specialists, and clinical experts can collaborate to modify them online, and the modification records are automatically archived to meet the traceability requirements of the ISO15189.

2. Scientific research laboratories

The personalization of scientific research applications is reflected in two dimensions: one is the information dimension (the specific parameters of the project need to be recorded), and the other is the process dimension (it needs to adapt to the stage changes of the project).

1) Custom templates

The LIS system of a cancer research institute splits the application template into basic modules + extended modules:

• Basic modules(Fixed field): project number, sample type, applicant, ethics approval number (required, the system will verify the validity of this number on the official website of the Ethics Committee);
• Extension modules(Optional components): If the circRNA sequencing template contains fields such as RNA Integrity Value (RIN) library concentration sequencing platform (NovaSeq/HiSeq), researchers can add/delete them by dragging and dropping, and can also set field attributes (e.g., sequencing depth is numeric + unit type, range is limited to 10X-500X).

More importantly, the version evolution function of the template: when the study moves from the sample screening stage to the validation stage and needs to add primer sequence fields, users can copy the old template and mark it with V2.0, and the system will automatically retain the history of V1.0, making it easy to trace the data differences at different stages.

2) Project binding

In the practice of antimicrobial peptide screening of the National Natural Science Foundation of China, the integration of LIS and the scientific research project management system is reflected in three levels:

1. Metadata association: Select the project number (such as 82371352) when applying, and the system will automatically fill in the project start and end time, person in charge, partner unit and other information to avoid manual entry errors.
2. Data attribution lock: After the test results are generated, they will be labeled as project-specific, and even if the sample is used for other topics in the future, the original result still belongs to the original project and cannot be modified (only secondary analysis notes can be added).
3. stage adaptation: In the transition from pre-experiment to formal experiment, the system will automatically adjust the verification rules – the sample size allowed in the pre-experiment stage is insufficient (due to the lack of samples in exploratory experiments), and the sample size is forced to be checked ≥ 3 times in the formal experiment.

2. Specimen acceptance

The core of specimen verification is to confirm identity + evaluation status, clinical scenarios are afraid of Zhang Guanli Dai, and scientific research scenarios are afraid of information faults.

1. Clinical Laboratories

At the verification desk of the laboratory department of a central hospital, 2,000 specimens are verified every day through the three-step process of scanning, comparing, and confirming, and each step hides detailed design:

1) Scanning code

Specimen tubes are printed with two codes at the same time:

• Clear code(Code128 code): contains patient ID + item code, which can be quickly identified by the code scanner;
• code(QR code): encrypted storage of collection time, collector, specimen tube type (such as EDTA anticoagulant tube), only LIS system can decrypt.

When verifying, the system first scans the clear code to retrieve the application information, and then scans the secret code to compare whether the specimen tube type matches the item (if the blood routine requires EDTA tubes, if it is scanned out as heparin tubes, it will indicate the wrong tube type). Through this design, a hospital reduced the error rate of specimens that did not match the application from 1.2% to 0.15%.

2) Unqualified specimens

The system constructs the whole process of discovery, feedback, rectification, and verification for unqualified specimens:

• found: When the inspector selects the reason for refusal, the system will pop up the subdivision options (such as hemolysis is divided into mild/moderate/severe, with a diagram);
• feedback: The message is sent to the collection nurse station through the hospital’s SMS platform, including the patient’s bed number + reason for refusal + rectification guide (if severe hemolysis needs to be re-collected, avoid violent shaking during collection);
• Rectification: After receiving it, the nurse station needs to click to confirm, and record the re-collection time, and the system automatically calculates the interval from rejection to re-collection (target ≤ 1 hour);
• verify: Generate the “Unqualified Specimen Analysis Report” every month, use the Pareto chart to show the top 3 reasons (such as the monthly collection volume is less than 60%), and the laboratory will cooperate with the nursing department for targeted training (such as making a schematic diagram of the specimen volume scale and pasting it on the collection table).

2. Scientific research laboratories

The acceptance of scientific research specimens is not a one-time confirmation, but a whole chain of archiving. A biopharmaceutical company’s LIS system generates a digital portfolio for each specimen, which contains three core modules:

1) Sample traceability module

Documenting the full path from source to lab:

• Clinical sample: enrollment criteria of associated patients (such as lung cancer stage III., no chemotherapy), informed consent number, validity period of GCP training certificate of collectors;
• Animal samples: including strain purity (such as SPF level certificate of C57BL/6 mice), feeding environment (temperature 22±1°C, humidity 50±5%), route of administration (intraperitoneal injection/gavage), etc.

When analyzing the efficacy data of hypoglycemic drugs, a team found that the age record of a group of mice was 8 weeks through this module, but the feeding log showed that it was actually 6 weeks, and the abnormal data caused by age deviation was eliminated in time.

2) Store tracking module

Deep linkage with IoT devices:

• The refrigerator/liquid nitrogen tank is equipped with RFID tags, and when the specimen is placed, the system reads the tag position through the antenna (such as the 5th grid on the 3rd floor of the liquid nitrogen tank area A), and records the liquid level (≥80%) when it is put in.
• When taking out, the system will automatically start a countdown reminder (if the RNA sample needs to be extracted within 30 minutes, the screen will display the dynamic progress bar for the remaining 15 minutes);
• When restoring, you need to take a photo of the remaining amount of the specimen tube and upload it, and the system will automatically compare the amount before and after removal (such as the original 1ml, take out 2ml, and the restoration should be 0.8ml) to avoid misoperation.

3. Inspection and implementation

Inspection and execution is the core link of data generation, and clinical research pursues less manual intervention, while scientific research needs to be flexibly adjusted.

1. Clinical Laboratories

Behind the unmanned assembly line in the biochemical inspection area of a hospital is the precise scheduling of equipment by the LIS system:

1) Equipment allocation

The system builds a device capability matrix that contains the following for each device:

• Project coverage (for example, AU5800 can do 12 liver functions, Cobas6000 can do 4 blood lipids);
• throughput per hour (e.g., 1200 tests per hour for the AU5800 and 800 tests for the Cobas6000);
• Current load (shows the number of queued samples in real time).

When the specimens are accepted, the system will be allocated like a dispatcher: if there is a sudden increase in liver function applications, the excess load will be automatically allocated to the spare Cobas8000 (although the processing speed is a little slower, but it avoids backlogs). This dynamic allocation reduced the median standard turnover time (TAT) for the department from 65 minutes to 40 minutes.

2) Handmade projects

For projects that cannot be automated, such as microbial culture, the system design interactive how-to guide:

• Step decomposition: Sputum culture is divided into 4 steps: inoculation→ culture→ staining→ identification, each step contains operation points + time window (such as entering the incubator within 30 minutes after inoculation);
• Multimedia assistance: key steps with video (such as Gram dyeing time control), click to play;
• Real-time recording: For each completed step, the inspector needs to scan his work card to confirm, and the system records the operator + time, and if the timeout is not turned on at the specified time, a reminder will be triggered.

2. Scientific research laboratories

The variability of scientific research experiments requires LIS systems to have low-code capabilities, which is reflected in the system design of a gene editing laboratory:

1) Customize the process

The researchers designed the experimental steps using a drag-and-drop editor interface similar to a flowchart:

• Basic steps (such as sample centrifugation reagent addition) can be dragged and dropped directly from the left component library;
• The conditional branch (e.g., diluted if the OD value is >8, otherwise it will be directly put on the machine) is set by the if-else node;
• Parameter associations (e.g., centrifugation time = sample volume ×0.5 min/ml) are implemented via the formula editor.

When designing the CRISPR editing efficiency detection process, a team streamlined the steps from 12 steps to 8 steps, and the newly enrolled graduate students could master the operation in just one hour.

2) Toolchain integration

Systems and scientific research tools to build a seamless data flow through train:

• Mass spectrometry data: After the Thermo .raw file is generated, LIS automatically calls MaxQuant’s API to pass in digestion parameters (such as Trypsin, 2 missed cleavage sites), and transmits the protein quantification results back to the system after the analysis is completed.
• Sequencing data: After the bcl file of Illumina is converted to fastq format through the BaseSpace platform, LIS triggers the BWA comparison process to store the mutation frequency with the sample information.
• Statistical analysis: Support direct call of R scripts, such as passing drug concentration and cell viability data into preset scripts, automatically generating IC50 curves, and calculating confidence intervals.

4. Review of results

Result review is a key link in removing falsification and preserving truth, with clinical focus on clinical matching and scientific research focusing on repeatable verification.

1. Clinical Laboratories

The dual-track review process of blood routine in a hospital combines the advantages of machine rapid screening and manual precise judgment:

1) Automatic review

Instead of a fixed reference range, the system’s automated audit rules are individualized dynamic thresholds:

• For old patients: refer to their past 3 historical results, if the deviation from the mean is < 20% and within the normal range, it will be automatically passed;
• For new patients: age/sex-specific reference ranges (e.g., the normal range of leukocytes in newborns is higher than that of adults);
• For critical values: Set graded responses, such as platelet <20×10⁹/L as the first-level critical value, the system will not only automatically intercept it, but also immediately push it to the doctor’s mobile app (and require confirmation within 10 minutes).

This design allows 85% of routine blood results to be automatically reviewed, and only 15% of abnormal samples to be processed manually.

2) Manual review

The system provides multi-dimensional visual data dashboards for auditing doctors to assist in decision-making:

• Trend chart: Show the changes in the results in the past 6 months (e.g., blood glucose from 1→6.3→7.8mmol/L, the curve is marked with an upward trend in red);
• Clinical association: the right side shows the patient’s medication records (e.g., taking prednisone in the past 7 days may lead to elevated white blood cells), diagnosis updates (e.g., adding ‘hyperthyroidism’);
• Operation Notes: Displays abnormalities in the inspection process (such as temporary calibration of the instrument channel 3, the result has been corrected).

2. Scientific research laboratories

The LIS system of a microbiology laboratory ensures trustworthy data through three-layer full-link traceability:

1) Data quality verification

The system automatically calculates the quality metrics of the experiment, from numerical values to more intuitive confidence:

• Repeatability: CV value of 3 replicates in the same experimental group (e.g., cell viability CV> 15% is marked in red);
• Accuracy: deviation from quality control (for example, ATCC standard strains are used to verify the antibacterial rate, a deviation of > 10% indicates that the reagent may fail);
• Completeness: Check for missing data (e.g., the result at the 5th time point of a sample is not entered).

When a team detected the antibacterial peptide antibacterial rate, the system suggested CV=22%, and it was traceably found that it was manual uneven sampling, and the CV dropped to 8% after re-dosing with the dispenser.

2) Traceability tree

The system uses samples as root nodes to generate a multi-branch traceability tree to visualize the root cause of abnormalities with one click:

• Reagent branch: record batch number, expiration date, batch quality inspection report (such as pH value 2±0.1 of a batch of LB medium), opening time;
• Instrument branch: including calibration records (such as the calibration certificate number of the microplate reader, the OD value of the calibration point), maintenance records (such as the use time of the ultraviolet lamp for 230 hours, the remaining 70 hours of life);
• Operation branch: operator + time + environmental parameters for each step (e.g., step 2 is completed in the biological safety cabinet, the temperature and humidity in the cabinet are 25°C/50%).

5. Report release

The report is the final product of the inspection process, which is available in a timely manner for clinical pursuit and in-depth support for scientific research, so that the value of data can be accurately reached.

1. Clinical Laboratories

The report publishing system of a hospital covers three types of users: doctor-patient-management, and each end is designed with its own focus, and multi-terminal collaboration makes the report accessible at any time:

1) Doctor’s end

The HIS system’s physician workstation has a built-in test result module:

• Click on patient information to view reports (no need to switch systems);
• Comparison of supporting results (such as juxtaposing the results of this and the last blood routine in a table, and the difference items are marked in yellow);
• Provide clinical interpretation (e.g., elevated creatinine may indicate renal function damage, it is recommended to combine it with routine urine examination), and the interpretation content is jointly prepared by the laboratory department and the nephrology department.

2) Patient side

Patients can easily access reports in three ways:

• WeChat official account: Receive the template message that the report has been issued, click to view the PDF version (including electronic signature), and support long press to save to the mobile phone;
• Self-service printer: After scanning the medical card, the system first displays a list of printable reports (avoid repeated printing), and there is a QR code at the bottom of the printed document (scanning the code to check the anti-counterfeiting information);
• Inpatient ward: The printer at the nurse’s station automatically prints a critical value report (e.g., elevated cardiac troponin) with management recommendations (e.g., immediate cardiology consultation).

2. Scientific research laboratories

The design of scientific research reports revolves around supporting the writing of papers, which require deeper integration, and the report of a university laboratory contains five core parts:

1) Abstract of experimental design

Automatic extraction of project information: project name (validation of circRNA as a marker of lung cancer) + sample size (100 patients/100 controls) + detection method (qPCR, primer sequence see attachment).

2) Raw data matrix

Export in Excel format and contains:

• sample information sheet (numbering, grouping, clinical characteristics);
• Test result table (Ct value, relative expression amount, internal reference correction value);
• Quality control data sheet (amplification efficiency, peak dissolution curve).

3) Statistical analysis results

Automatic generation:

• Differences between groups (P-value, mean, ± standard deviation of t-test);
• Correlation analysis (Pearson correlation coefficient r and P value, with scatter plot);
• ROC curve (used to evaluate the diagnostic efficacy of markers, AUC value is automatically calculated).

4) Outlier description

Label all data that deviate from expectations (e.g., sample ID-108 has a Ct value of >35, may be due to RNA degradation and has been eliminated) and include the rationale for exclusion.

5) Permission control

Access via Project Key Management:

• Members in the project team can enter the key to download all the data.
• Journal reviewers can only view statistical results + charts to hide patient privacy information;
• All access records (e.g., 2023-10-01, a journal’s editorial office IP download chart) are automatically archived to comply with HIPAA compliance requirements.

In the future, with the integration of AI (such as using machine learning to predict TAT time) and digital twins (such as virtual simulation laboratory process optimization) and other technologies, the functional boundaries of LIS will continue to expand, but the design logic from and from the scene will not change. Only by always rooted in the real pain points of the laboratory can LIS truly become the invisible wings of medical and scientific research.

(Note: The cases in this article are all from real practice, and the names and data involved have been desensitized)