Accurate reference intervals are crucial for disease diagnosis and health assessment in clinical laboratories. The traditional direct method is limited by its high cost and complex procedures，and indirect method leverages real-world data to establish reference intervals cost-effectively，offering greater potential for widespread adoption. Methodologically，the field has progressed from traditional graphical techniques reliant on manual interpretation （Hoffman and Bhattacharya） to robust statistical models （KOSMIC and refineR），and further to the integration of machine learning algorithms （mixture density networks and deep learning），enabling intelligent and automated estimation of the healthy population parameter distribution from mixed datasets. In application，the capacity of indirect methods to process vast datasets has successfully catalyzed a shift from generic reference interval towards precise stratification，effectively revealing and quantifying the specific influences of factors like age，sex and geography on physiological markers. Latest methods now transcend fixed partitions，utilizing techniques such as generalized additive models for location，scale and shape and quantile regression to construct dynamic reference curves that vary continuously with age. Furthermore，they facilitate the development of personalized reference interval by incorporating within-subject biological variation and longitudinal data. Despite persistent challenges related to data quality，model selection and clinical validation，the synergy between indirect methods and big data is unequivocally propelling laboratory medicine toward a more precise and personalized development.
Large language models （LLM） such as DeepSeek and ChatGPT provide an intelligent foundation for laboratory information ecosystem，facilitating the realization of intelligent laboratory medicine. At present，LLM still have limitations，such as the lack of application scenario implementation solutions，the absence of full life cycle management and no fine-tuning for laboratory medicine. This review introduces the technical solutions and expected effects of LLM in application scenarios，such as pre-，intra- and post-laboratory determination and laboratory management，constructs a management framework for the full life cycle of LLM，and realizes the deep integration of LLM and laboratory medicine through technologies，such as prompt engineering，retrieval augmented generation and fine-tuning. This review expounds the current application characteristics，challenges and countermeasures of LLM as well，providing a reference for further promoting the development of intelligent laboratory medicine.
Objective To investigate the consistency of different algorithms for establishing the reference range of serum cystatin C（Cys C）by indirect method. Methods Healthy subjects >18 years old who underwent serum Cys C determination at the First Affiliated Hospital of Zhejiang University School of Medicine were enrolled. The 32 266 healthy subjects from January 2019 to December 2023 were used as establishment set，and the 6 531 healthy subjects from January to June 2024 were used as validation set. Harris & Boyd method and the GAMLSS model were used to determine whether to group by gender or age. Six indirect method algorithms, {non-parametric method [the Clinical and Laboratory Standards Institute（CLSI）EP28-NP document]，parametric method（CLSI EP28-P document），TMC，refineR，Kosmic and Hoffmann}, were used to establish 95% reference ranges for each group，and the bias ratio （BR） of different algorithms was compared. The reference ranges established by different algorithms were verified. Results The serum Cys C level in males was higher than that in females （P<0.01），and the serum Cys C levels among different age groups were statistically significant （P<0.01）. Based on the reference range established by non-parametric method as the reference（the reference range of serum Cys C for males was：0.61-1.01 mg·L-1 for 19-39 years old，0.63-1.10 mg·L-1 for 40-60 years old，and 0.69-1.30 mg·L-1 for≥61 years old；the reference range of serum Cys C for females was：0.50-0.85 mg·L-1 for 19-46 years old，and 0.56-1.05 mg·L-1 for≥47 years old），the reference ranges established by parametric method，refineR and Hoffmann were basically consistent （|BR|≤0.330），while there were significant differences （|BR|≥0.445） with TMC（for males 19-39 years old，and for females overall）and Kosmic（for males overall and 19-39 years old，and for females overall）. In the validation set，the pass rates of the reference ranges established by the 6 algorithms for males and females were 90.0%-97.3% and 90.5%-93.0%，respectively. The reference range of serum Cys C was age-related，showing a downward trend starting from 19 years old，gradually increasing for males starting from 40 years old，and gradually increasing for females starting from 47 years old. Conclusions There are certain differences in the reference ranges of serum Cys C established by the 6 indirect method algorithms in Hangzhou. Among them，non-parametric method and refineR algorithms have simpler operations and more stable results. The reference range of serum Cys C for males in Hangzhou is higher than that for females，and it is non-linearly related to age.
Objective To establish the reference ranges for alpha-fetoprotein（AFP），carcinoembryonic antigen（CEA），carbohydrate antigen 72-4（CA72-4）and carbohydrate antigen 19-9（CA19-9）in Urumqi region by indirect method. Methods The data of AFP，CEA，CA72-4 and CA19-9 from laboratory information system（LIS）of Xinjiang Production and Construction Corps Hospital from January 2021 to July 2024 were collected. Outliers were removed using the Tukey method. The standard deviation ratio（SDR）and deviation ratio were calculated based on multiple-factor linear regression and variance component models to determine whether separate reference ranges should be established by gender and age. The 95% reference ranges for each indicator were established using non-parametric methods. The relative deviation between the upper limit of each indicator reference range and the reference range established by direct method was calculated，compared to reference change value（RCV）. Results The SDR of AFP was <0.3，and the deviation ratio was 0.489，indicating that separate reference ranges should be established by gender （male<7.55 ng·mL-1，female<6.67 ng·mL-1）. The SDR of CEA by gender and age were 0.324 and 0.553，respectively，indicating that separate reference ranges should be established by gender and age （male<60 years old<4.54 ng·mL-1，≥60 years old<5.85 ng·mL-1；female<60 years old < 2.89 ng·mL-1，≥60 years old <5.65 ng·mL-1）. The reference ranges of CA72-4 and CA19-9 were <13.7 and <24.85 U·mL-1，respectively. The upper limits of the self-established reference ranges of AFP，CEA，CA72-4 and CA19-9 were within the relative deviation of RCV from the reference ranges established by direct method. Conclusions The reference ranges for AFP，CEA，CA72-4 and CA19-9 established in Urumqi region using indirect method have good consistency with those established by direct method，which can provide more appropriate reference ranges for clinical decision-making，suitable for different people.
Objective To establish the reference range of carbohydrate antigen 50 （CA50 ）for adults in Shanghai by indirect method （Hoffmann method） and verify its applicability. Methods The 46 764 cases of original CA50 determination data in the laboratory information management system of Zhongshan Hospital of Fudan University from August 2022 to July 2023 were collected for the establishment of CA50 reference range. Totally，691 healthy subjects at Zhongshan Hospital of Fudan University from January to March 2024 were enrolled for the verification of the self-established reference range. The individuals from the physical examination population of Zhongshan Hospital of Fudan University from August 2022 to March 2024 with CA50≥25 U·mL-1（current reference range）but lower than the self-established reference range were selected to verify the applicability of the self-established reference range. Turkey method was used to eliminate outliers. Z test was used to compare whether there were differences in CA50 levels among different genders and ages. The 95% reference range of serum CA50 was established using the Hoffmann method. Results Totally，17 773 valid data were obtained. The Z test results showed that there was statistical significance in CA50 levels among different genders and ages（Z value > Z* value）. The 95% reference ranges of CA50 were established for different genders and ages using the Hoffmann method. That for male individuals aged 18-59 years was <27.30 U·mL-1，and that for female individuals was <29.62 U·mL-1. That for male aged ≥60 years was <36.85 U·mL-1，and that for female individuals was <34.85 U·mL-1. The CA50 levels of 691 healthy individuals were within the self-established reference range，and the reference range verification was successful. Among the 191 individuals with CA50≥25 U·mL-1（current reference range）and lower than the self-established reference range，38（19.90%）individuals had abnormal liver function，renal function，B-ultrasound or computed tomography（CT）results，while the other 153 （80.10%） individuals had normal examination results. Conclusions The Hoffmann method has established the reference range of CA50 for adults in Shanghai by gender and age，which is different from the current reference range and is more applicable in clinical practice.
Objective To establish the reference ranges for the 4 lipid parameters of adults in Lhasa by indirect method，and to provide a reference for the prevention and treatment of lipid disorders in the population of the plateau region. Methods Healthy individuals aged 18-80 years from Lhasa People's Hospital from October 2023 to October 2024 were enrolled. Serum total cholesterol（TC），triglyceride（TG），low-density lipoprotein cholesterol（LDL-C）and high-density lipoprotein cholesterol（HDL-C）were determined. The groups were classified by gender and age（18-29 years old，30-39 years old，40-49 years old，50-59 years old，60-69 years old，70-80 years old）. The original data were normalized，and outliers were removed using the Turkey method. The 95% reference ranges for the 4 lipid parameters were established using the Hoffmann method. Results The serum TG，TC and LDL-C levels in males were higher than those in females（P<0.001），while the serum HDL-C level was lower in males（P<0.001）. There was statistical significance in serum TG，TC and LDL-C levels between 18-29 years old group，30-39 years old group and 40-80 years old groups（P<0.001），while there was no statistical significance among 40-80 years old groups（P>0.05）. For females，the differences in serum TG and LDL-C levels were not statistically significant except for 60-69 years old group and 70-80 years old group（P>0.05），while the differences were statistically significant among the other age groups（P<0.05）. The serum TC levels showed no statistical significance among 50-59，60-69 and 70-80 years old groups （P>0.05），while significant differences were observed among the other age groups （P<0.05）. There was no statistical significance in serum HDL-C levels among males and females in each age group（P>0.05）. The reference ranges for TG，TC and LDL-C were established for males in 3 age groups（18-29 years old，30-39 years old and 40-80 years old），while the reference ranges for TG and LDL-C were established for females in 5 age groups（18-29 years old，30-39 years old，40-49 years old，50-59 years old and 60-80 years old）. The reference ranges for TC were established for females in 4 age groups（18-29 years old，30-39 years old，40-49 years old and 50-80 years old）. HDL-C reference ranges were established for males and females. The relative deviations between the self-established reference ranges and the current reference ranges for the 4 lipid parameters were calculated，and there were differences of varying degrees with respect to the reference change value（RCV）of adults item. Conclusions There are differences of varying degrees between the reference ranges of the 4 lipid parameters for adults in the plateau region and the current reference ranges. Appropriate reference ranges for lipid parameters should be established for different regions and populations.
Objective To prelimiarily establish the reference range of cytokeratin 19 （CYFRA21-1） by electro-chemiluminescence method and verify its applicability and the correlation with other relevant determination indicators. Methods A total of 1 642 apparently healthy subjects from Shanghai Municipal Hospital of Traditional Chinese Medicine of Shanghai University of Traditional Chinese Medicine from February 2023 to May 2024 were enrolled，and they were classified according to gender and age （≤40 years old，41-60 years old，≥61 years old）. Another 120 apparently healthy subjects who underwent routine physical examinations at the Shanghai Municipal Hospital of Traditional Chinese Medicine of Shanghai University of Traditional Chinese Medicine in May 2025 were enrolled for validating the self-established reference range. The serum CYFRA21-1 levels were determined by electro-chemiluminescence method，and the general data（gender，age）and laboratory indicator [carcinoembryonic antigen（CEA），alpha-fetoprotein（AFP），alanine aminotransferase（ALT），serum creatinine（SCr），fasting plasma glucose（FPG），total cholesterol（TC），triglyceride（TG）and white blood cell（WBC） count] results were collected. Spearman correlation analysis and partial correlation analysis were used to evaluate the correlation between CYFRA21-1 and other indicators. The reference range establishment procedure followed the requirements of the Clinical and Laboratory Standards Institute（CLSI）EP28-A3c document. Results Except for the difference in serum CYFRA21-1 levels between females in the ≤40-year-old group and those in the 41-60-year-old group（P>0.05），there was statistical significance in serum CYFRA21-1 levels among the other age groups of the same gender（P<0.001）. The serum CYFRA21-1 level of males in the 41-60-year-old group was higher than that of females（P=0.005），and there was no statistical significance between males and females in the ≤40-year-old group and ≥61-year-old group（P>0.05）. The upper limit of the reference range was set at the 95th percentile，and the reference range of CYFRA21-1 was established by age group：≤40 years old ≤3.47 ng·mL-1，41-60 years old ≤4.08 ng·mL-1，≥61 years old≤4.97 ng·mL-1. After adjusting for age，CYFRA21-1 was positively correlated with CEA，SCr，FPG and WBC count （r values were 0.114，0.198，0.160 and 0.088，respectively，P<0.001）.Among the 120 apparently healthy subjects used for validating the reference range，the number of cases exceeding the reference range for those≤40 years old，40-60 years old and≥61 years old were 1（2.5%），1（2.5%）and 2（5.0%），respectively. The self-established reference range validation was successful. Conclusions The serum CYFRA21-1 level shows an increasing trend with age，and the reference range of CYFRA21-1 for different age groups in Shanghai has been preliminarily established.
Objective To evaluate the interpretation ability of different large language models（LLM）for tumor marker determination reports and provide a reference for the clinical application of LLM. Methods The relevant data of patients who underwent tumor marker determination at the First Affiliated Hospital of Zhejiang University School of Medicine in 2024 were collected. Stratified random sampling was conducted using the sampling package of R software，and 200 determination reports were randomly selected. DeepSeek R1，Qwen 3，KIMI and ChatGPT 4.1 were used to interpret 200 determination reports. Totally，2 junior and 2 senior assessors were selected to evaluate the interpretation quality of 4 LLM on a 10-point scale. The differences in interpretation ability among different LLM were evaluated using the Friedman and Wilcoxon tests. Results The overall scores of the tumor marker determination reports that could be interpreted by all 4 LLM from high to low were DeepSeek R1[9（8，10）points]，Qwen 3[9（8，10）points]，KIMI[8（6，10）points] and ChatGPT 4.1 [7（5，9）points]（P<0.001）. DeepSeek R1 had 0.3%，Qwen 3 had 3.6%，KIMI had 19.0%，and ChatGPT had 27.2% with a score of ≤5 points. There was statistical significance in the score given by assessors of different levels for the 3 LLM except for DeepSeek R1（P<0.001）. The scores for the comprehensiveness，accuracy，clarity and relevance of the LLM report interpretation were consistently good（P<0.001）. The interpretation ability of LLM and the ability to identify abnormal indicators were superior to the ability to analyze the causes of abnormalities and the ability to make clinical recommendations. The incidences of hallucinations in DeepSeek R1 and Qwen 3 were 3.0% and 2.5%，which were lower than those in KIMI（13.0%）and ChatGPT 4.1（16.0%）. Conclusions LLM report interpretation can be used as an auxiliary tool，but there are performance differences among different models. In practical applications，strict supervision should be carried out，and continuous improvement should be made to reduce the risk of incorrect report interpretation.
Objective To establish a candidate reference measurement procedure for serum oxcarbazepine （OXC） and 10-hydroxy carbamazepine （MHD） using isotope dilution liquid chromatography-tandem mass spectrometry （ID LC-MS/MS），and to evaluate its performance. Methods The samples were pre-treated using protein precipitation （PPT） with OXC-2H4 and MHD-2H4 as internal standards. The chromatography used a Phenomenex Kinetex F5 column，with a mobile phase of 0.1% formic acid water -0.1% formic acid methanol and gradient elution；the mass spectrometry used positive ion multiple reaction monitoring （MRM） mode. The linearity，limit of quantification （LOQ），matrix effect，accuracy，precision，stability and uncertainty of the method were evaluated，which was compared with the routine determination system （LC-MS/MS） in clinical laboratories. Results The linear range of ID LC-MS/MS for serum OXC and MHD was 0.02-50.00 μg·mL-1，and the LOQ was 0.015 μg·mL-1. The relative matrix effect of OXC was -2.09%-2.12%，and the recovery rate was 100.00%-101.62%. The relative matrix effect of MHD was -0.87%-2.99%，and the recovery rate was 101.39%-102.08%. The within-run and between-run coefficients of variation（CV）of OXC and MHD were both <3%. The relative expanded uncertainty was ≤2.57%. The results of the 2 methods for determining serum OXC and MHD had good correlation. Using the routine determination system calibrated by the candidate reference measurement procedure to determine 5 serum samples with different concentrations from the 2024 National Serum Therapeutic Drug Monitoring Inter-laboratory Quality Evaluation Program，the deviation was 0.00%-3.64%. Conclusions A candidate reference measurement procedure for serum OXC and MHD based on ID LC-MS/MS was successfully established，which can be used for the standardization of OXC and MHD therapeutic drug monitoring （TDM） projects.
Objective To investigate the correlation between serum neuregulin-1（NRG-1） and matrix metalloproteinase-9（MMP-9）levels and the cognitive function of patients with stable schizophrenia. Methods A total of 104 patients with stable schizophrenia（case group）and 70 healthy subjects（healthy control group）were enrolled from Tongji University Affiliated Mental Health Center from October 2023 to September 2024. The serum NRG-1 and MMP-9 levels were determined，and their cognitive functions were evaluated. Pearson correlation analysis was used to assess the correlation between serum NRG-1，MMP-9 levels and cognitive function. Logistic regression analysis was used to evaluate the influencing factors of cognitive dysfunction. Receiver operating characteristic（ROC）curve was used to evaluate the efficacy of NRG-1 and MMP-9 in determining the occurrence of cognitive dysfunction. Results Compared with healthy control group，the cognitive function of case group was lower（P<0.001），and the serum NRG-1 level was decreased（P<0.001），while the MMP-9 level was increased（P<0.001）. The cognitive function of healthy control group was not correlated with NRG-1 and MMP-9（P>0.05）. The cognitive function of case group [total score of repeatable battery for the assessment of neuropsychological status（RBANS）scale，immediate memory，verbal function，visual span，attention，delayed memory] was positively correlated with NRG-1（r values were 0.946，0.878，0.853，0.811，0.842 and 0.717，P<0.05）and negatively correlated with MMP-9（r values =-0.339，-0.372，-0.355，-0.072，-0.280 and -0.372，P<0.05）. Decreased NRG-1 level and increased MMP-9 level were risk factors for the occurrence of stable schizophrenia（P<0.001）. The areas under curves（AUC）of NRG-1，MMP-9 single and combined determinations for judging cognitive dysfunction were 0.787，0.723 and 0.858，respectively. Conclusions The serum NRG-1 and MMP-9 levels of patients with stable schizophrenia are related to cognitive function，and they have certain value in predicting cognitive dysfunction.
Objective To investigate the relationship between serum homocysteine（Hcy），S100B protein levels and cerebral metabolite indicators [N-acetylaspartate（NAA）/creatine ratio，choline complex（Cho）/creatine ratio] and cognitive impairment in elderly patients with acute ischemic stroke（AIS）. Methods A total of 92 elderly AIS patients from Shanghai Huangpu Xiangshan Hospital of Traditional Chinese Medicine from January 2021 to June 2023 were enrolled. The clinical data at admission were collected，and hydrogen proton magnetic resonance spectroscopy was used to determine cerebral metabolites [NAA，Cho，creatine]，and the NAA/creatine ratio and Cho/creatine ratio were calculated. Laboratory indicators [Hcy，S100B protein，total cholesterol（TC），triglyceride（TG），low-density lipoprotein cholesterol（LDL-C），high-density lipoprotein cholesterol（HDL-C），interleukin （IL）-1β，IL-6，tumor necrosis factor-alpha（TNF-α），IL-8，IL-10] were determined. All the patients were followed up for 6 months，and they were classified into cognitive impairment group（46 cases）and non-cognitive impairment group（46 cases）according to the Montreal Cognitive Assessment（MoCA）. Pearson correlation analysis was used to evaluate the correlation between each index and the MoCA score. Results The NAA/creatine ratio in the left and right hemispheres of the brain in cognitive impairment group was lower than that in non-cognitive impairment group（P<0.001），and the Cho/creatine ratio was higher（P<0.01）. The serum Hcy and S100B protein levels in cognitive impairment group were higher than those in non-cognitive impairment group（P<0.05）. There was no statistical significance in serum IL-1β，IL-6，TNF-α，IL-8，IL-10，TC，TG，LDL-C and HDL-C levels between the 2 groups（P>0.05）. The NAA/creatine ratio in the left and right hemispheres of the brain in AIS patients was negatively correlated with the MoCA score（r values were -0.13 and -0.20，respectively，P<0.05），and the Cho/creatine ratio was positively correlated with the MoCA score（r values were 0.17 and 0.14，respectively，P<0.05）. Serum Hcy and S100B protein were positively correlated with the MoCA score（r values were 0.47 and 0.21，respectively，P<0.05）. Conclusions The NAA/creatine ratio，Cho/creatine ratio and serum Hcy and S100B protein levels in the brain are related to cognitive impairment in AIS patients and may serve as the biological indicators for assessing the risk of cognitive impairment.
Renal cancer is a common tumor in the urinary system. Exosomes participate in various aspects of tumor occurrence and development，such as intercellular communication，immune escape of tumor cells，angiogenesis，metastasis and drug resistance. Exosomes play an important role and have a promising clinical application prospect in the early diagnosis，treatment and prognosis assessment of renal cancer. This review introduces the methods for isolating，extracting and identifying exosomes，elaborates on the role and mechanism of exosomes in renal cancer and their functions in the diagnosis，treatment and prognosis of renal cancer.
Inductively coupled plasma mass spectrometry （ICP-MS） has been widely applied in environmental monitoring，food，pharmaceuticals，public health and so on. In the recent years，ICP-MS has begun to be promoted and applied in the field of clinical laboratory medicine. This review focuses on the application of ICP-MS in clinical laboratory medicine，including clinical determination，standardization and reference methods，as well as other research aspects，and it provides a future outlook.
