Measuring the performance of an artificial intelligence-based robot that classifies blood tubes and performs quality control in terms of preanalytical errors: A preliminary study.

OBJECTIVES: Artificial intelligence-based robotic systems are increasingly used in medical laboratories. This study aimed to test the performance of KANKA (Labenko), a stand-alone, artificial intelligence-based robot that performs sorting and preanalytical quality control of blood tubes. METHODS: KANKA is designed to perform preanalytical quality control with respect to error control and preanalytical sorting of blood tubes. To detect sorting errors and preanalytical inappropriateness within the routine work of the laboratory, a total of 1000 blood tubes were presented to the KANKA robot in 7 scenarios. These scenarios encompassed various days and runs, with 5 repetitions each, resulting in a total of 5000 instances of sorting and detection of preanalytical errors. As the gold standard, 2 experts working in the same laboratory identified and recorded the correct sorting and preanalytical errors. The success rate of KANKA was calculated for both the accurate tubes and those tubes with inappropriate identification. RESULTS: KANKA achieved an overall accuracy rate of 99.98% and 100% in detecting tubes with preanalytical errors. It was found that KANKA can perform the control and sorting of 311 blood tubes per hour in terms of preanalytical errors. CONCLUSIONS: KANKA categorizes and records problem-free tubes according to laboratory subunits while identifying and classifying tubes with preanalytical inappropriateness into the correct error sections. As a blood acceptance and tube sorting system, KANKA has the potential to save labor and enhance the quality of the preanalytical process.

Serum Level of Biotin Rather than the Daily Dosage Is the Main Determinant of Interference on Thyroid Function Assays.

INTRODUCTION: Several case reports stress that high-dose biotin causes incorrect laboratory results. However, the extent of this interference in children is not systematically studied. AIM: To assess factors associated with biotin interference on thyroid function tests in subjects with biotinidase deficiency. METHOD: The study included 44 children who were treated with oral biotin (Group 1, median dose: 10 mg/day [25-75p; 10-10], age: 1.83 years [1.04-2.90]) and 30 healthy subjects (Group 2, age: 1.05 years [0.37-3.37]). Thyroid function tests were performed with two different assays, and streptavidin-coated particles were used in order to remove biotin from serum samples of cases with biotin interference. RESULTS: The measurements were first performed with Beckman Coulter. In Group 1, remarkably high levels of fT3 and fT4 were found in 26 (59.1%) and 25 (56.8%) patients, respectively. Thyroid hormone functions were all normal in Group 2. Significantly higher serum biotin levels were detected in interference-positive children (p < 0.001). The serum biotin levels in Group 1 showed a strong positive correlation with fT3 (r = 0.867, p < 0.001) and fT4 levels (r = 0.905, p < 0.001). A serum biotin level of 80.35 µg/L was found to be the best cut-off value for predicting interference (sensitivity: 96.2% and specificity: 94.4%). When analyzed with Siemens Advia Centaur XP, all thyroid function tests were normal in both groups except in one patient (2.27%) with slightly elevated fT3 level in Group 1. Repeated tests with Beckman Coulter after neutralization of biotin with streptavidin magnetic particles in serum samples of the interference-positive cases revealed normal thyroid hormone levels. CONCLUSION: Interference is an important problem in thyroid function tests in nearly 60% of all children receiving biotin treatment for biotinidase deficiency. Serum levels of biotin rather than the dosage are the main determinant of interference, which can be eliminated by choosing appropriate laboratory methods.

Do serum BDNF levels vary in self-harm behavior among adolescents and are they correlated with traumatic experiences?

The aim of this study was to compare serum brain-derived neurotrophic factor (BDNF) levels between adolescents that harm themselves, those that receive psychiatric treatment but do not harm themselves, healthy adolescents, and childhood traumas and to investigate the relationship between traumatic experiences and serum BDNF levels. The cases were divided into two groups of 40 adolescents exhibiting self-harm behavior (self-harm/diagnosed group) and 30 adolescents receiving psychiatric treatment but not exhibiting self-harm behaviors (non self-harm/diagnosed group). The control group (healthy control group) consisted of 35 healthy adolescents with no psychiatric disorders or self-harm behaviors. The adolescents were asked to fill in the Inventory of Statements About Self Injury (ISAS) and Childhood Trauma Questionnaire (CTQ). For BDNF measurement, blood samples were taken from the cases and controls. The serum BDNF level of self-harming adolescents who used the self-cutting method was significantly lower than that of other groups, and serum BDNF levels decreased with the increase in the emotional neglect and abuse severity of self-harming adolescents during childhood. In our study, serum BDNF levels decreased with the increase in emotional abuse in self-harming adolescents. This finding may indicate that neuroplasticity can be affected by a negative emotional environment during the early period.

Artificial Neural Network Approach in Laboratory Test Reporting: Learning Algorithms.

OBJECTIVES: In the field of laboratory medicine, minimizing errors and establishing standardization is only possible by predefined processes. The aim of this study was to build an experimental decision algorithm model open to improvement that would efficiently and rapidly evaluate the results of biochemical tests with critical values by evaluating multiple factors concurrently. METHODS: The experimental model was built by Weka software (Weka, Waikato, New Zealand) based on the artificial neural network method. Data were received from Dokuz Eylül University Central Laboratory. "Training sets" were developed for our experimental model to teach the evaluation criteria. After training the system, "test sets" developed for different conditions were used to statistically assess the validity of the model. RESULTS: After developing the decision algorithm with three iterations of training, no result was verified that was refused by the laboratory specialist. The sensitivity of the model was 91% and specificity was 100%. The estimated κ score was 0.950. CONCLUSIONS: This is the first study based on an artificial neural network to build an experimental assessment and decision algorithm model. By integrating our trained algorithm model into a laboratory information system, it may be possible to reduce employees' workload without compromising patient safety.