Failure state prediction for automated analyzers for analyzing a biological sample

1. A computer-implemented method for predicting a failure state of an automated analyzer for analyzing a biological sample, the method comprising:

a determining, by the automated analyzer, a machine learning prediction algorithm for predicting a failure state of an automated analyzer based on historic analyzer data, wherein the historic analyzer data includes historic data regarding the occurrence of failure states and historic calibration data and/or quality control data, and wherein the machine learning prediction algorithm is configured to predict a failure state of the automated analyzer based on calibration data and/or quality control data generated by an automated analyzer;

obtaining, by the automated analyzer, calibration data and/or quality control data of the automated analyzer; and

processing, by the automated analyzer, the calibration data and/or quality control data by using the machine learning prediction algorithm to autonomously predict a failure state of the automated analyzer,

wherein determining, by the automated analyzer, the machine learning prediction algorithm includes training an initial machine learning prediction algorithm using the historic data regarding the occurrence of failure states and the historic calibration data and/or quality control data,

wherein the machine learning prediction algorithm is configured to have one output state indicating that a failure state is expected and a second output state indicating that no failure state is expected,

wherein determining, by the automated analyzer, the machine learning prediction algorithm comprises determining a length of a reactive interval between a period of time in which the quality control and/or the calibration data is used to predict a failure event and the failure event.

2. The computer-implemented method of claim 1 , wherein the historic analyzer data includes historic data retrieved from a plurality of automated analyzers sharing one or more characteristics with the automated analyzer.

3. The computer-implemented method of claim 2 , wherein the historic data regarding the occurrence of failure states includes service data of the plurality of automated analyzers.

4. The computer-implemented method of claim 1 , further comprising finding, by the automated analyzer, a relationship between the historic data regarding the occurrence of failure states and the historic calibration data and/or quality control data to determine the machine learning prediction algorithm.

5. The computer-implemented method of claim 4 , further comprising:

classifying, by the automated analyzer, the historic data regarding the occurrence of failure states by a classification technique; and

determining, by the automated analyzer, a relationship between the historic calibration data and/or quality control data.

6. The computer-implemented method of claim 1 , further comprising solving, by the automated analyzer, a binary classification problem to determine the machine learning prediction algorithm.

7. The computer-implemented method of claim 6 , wherein the failure state is one of: a failure state requiring an emergency service with a visit at the automated analyzer, a failure state requiring an intervention by an operator of the automated analyzer, or a failure of a component of the automated analyzer.

8. The computer-implemented method of claim 1 , further comprising averaging, by the automated analyzer, the calibration data and/or quality control data over a plurality of assays available on the automated analyzer to determine the machine learning prediction algorithm.

9. The computer-implemented method of claim 1 , further comprising:

calculating, by the automated analyzer, distribution information for numeric calibration data and/or quality control data; and

generating, by the automated analyzer, the machine learning prediction algorithm based on the calculated data.

10. The computer-implemented methods of claim 1 , further comprising:

calculating, by the automated analyzer, frequency information for categorical calibration data and/or quality control data; and

generating, by the automated analyzer, the machine learning prediction algorithm based on the calculated data.

11. The computer-implemented method of claim 1 , further comprising:

determining a response signal of the automated analyzer to one or more standard calibrators having a known composition;

generating, by the automated analyzer, a relationship between the response signal and the known composition; and

including, by the automated analyzer, the relationship into the calibration data.

12. The computer-implemented method of claim 1 , further comprising:

determining, by the automated analyzer, a response signal of the automated analyzer to one or more one or more control materials with known target measurement values;

checking, by the automated analyzer, that the automated analyzer operates within a predetermined limit of accuracy and/or precision; and

including, by the automated analyzer, results of the checking step in the quality control data.

13. The computer-implemented method of claim 1 , further comprising continuously updating, by the automated analyzer, the machine learning prediction algorithm based on newly received calibration data and/or quality control data of the automated analyzer during operation of the automated analyzer.

14. The computer-implemented method of claim 1 , wherein determining the machine learning prediction algorithm comprises determining a machine learning prediction algorithm installed on the automated analyzer.

15. An automated analyzer for analyzing a biological sample, the automated analyzer comprising:

a detection unit to detect one or more properties of the biological sample;

a processor; and

a memory comprising a plurality of instructions stored thereon that, in response to execution by the processor, causes the automated analyzer to:

determine a machine learning prediction algorithm for predicting a failure state of an automated analyzer based on historic analyzer data, wherein the historic analyzer data includes historic data regarding the occurrence of failure states and historic calibration data and/or quality control data, and wherein the machine learning prediction algorithm is configured to predict a failure state of the automated analyzer based on calibration data and/or quality control data generated by an automated analyzer;

obtain calibration data and/or quality control data of the automated analyzer; and

process the calibration data and/or quality control data by using the machine learning prediction algorithm to autonomously predict a failure state of the automated analyzer,

wherein to determine the machine learning prediction algorithm includes to train an initial machine learning prediction algorithm using the historic data regarding the occurrence of failure states and the historic calibration data and/or quality control data,

wherein the machine learning prediction algorithm is configured to have one output state indicating that a failure state is expected and a second output state indicating that no failure state is expected,

wherein to determine the machine learning prediction algorithm further includes to determine a length of a reactive interval between a period of time in which the quality control and/or the calibration data is used to predict a failure event and the failure event.

16. The automated analyzer of claim 15 , wherein the plurality of instructions further causes the automated analyzer to:

find a relationship between the historic data regarding the occurrence of failure states and the historic calibration data and/or quality control data to determine the machine learning prediction algorithm;

classify the historic data regarding the occurrence of failure states by a classification technique; and

determine a relationship between the historic calibration data and/or quality control data.

17. The automated analyzer of claim 15 , wherein the failure state is one of: a failure state requiring an emergency service with a visit at the automated analyzer, a failure state requiring an intervention by an operator of the automated analyzer, or a failure of a component of the automated analyzer.

18. The automated analyzer of claim 15 , wherein the memory further comprises the machine learning prediction algorithm installed thereon; and

wherein to determine the machine learning prediction algorithm comprises to determine the machine learning prediction algorithm installed on the automated analyzer.

19. The automated analyzer of claim 15 , wherein the plurality of instructions further causes the automated analyzer to continuously update the machine learning prediction algorithm based on newly received calibration data and/or quality control data of the automated analyzer during operation of the automated analyzer.

20. One or more non-transitory machine-readable storage media comprising a plurality of instructions stored therein that, in response to execution by a processor, causes an automated analyzer to:

determine a machine learning prediction algorithm for predicting a failure state of an automated analyzer based on historic analyzer data, wherein the historic analyzer data includes historic data regarding the occurrence of failure states and historic calibration data and/or quality control data, and wherein the machine learning prediction algorithm is configured to predict a failure state of the automated analyzer based on calibration data and/or quality control data generated by an automated analyzer;

obtain calibration data and/or quality control data of the automated analyzer; and

process the calibration data and/or quality control data by using the machine learning prediction algorithm to autonomously predict a failure state of the automated analyzer,

wherein to determine the machine learning prediction algorithm includes to train an initial machine learning prediction algorithm using the historic data regarding the occurrence of failure states and the historic calibration data and/or quality control data,

wherein the machine learning prediction algorithm is configured to have one output state indicating that a failure state is expected and a second output state indicating that no failure state is expected,

wherein to determine the machine learning prediction algorithm further includes to determine a length of a reactive interval between a period of time in which the quality control and/or the calibration data is used to predict a failure event and the failure event.