Detection of cognitive impairment

A method of training first and second detection models for use in calculating test variables associated with first and second neuropsychological tests for detecting cognitive impairment of a patient, the method comprising:

• for each of a plurality of patients:
• conducting the first and second neuropsychological tests on the patient, wherein each of the first and second neuropsychological tests comprises displaying predetermined visual information or providing predetermined audible information prompting the patient to complete the test by making utterances;
• receiving audio data representing recorded utterances made by the patient during the tests;
• obtaining a first diagnosis made by a clinical practitioner listening to the recorded utterances made by the patient during the first neuropsychological test;
• obtaining a second diagnosis made by the clinical practitioner listening to the recorded utterances made by the patient during the second neuropsychological test;
• processing the audio data using a speech-to-text engine to produce a text transcription of the recorded utterances;
• processing the text transcription to calculate a first plurality of test variables associated with the first neuropsychological test;
• processing the text transcription to calculate a second plurality of test variables associated with the second neuropsychological test;
• associating the first plurality of test variables with the clinical practitioner's first diagnosis for the patient;
• training the first detection model using the first pluralities of test variables and the plurality of associated first diagnoses;
• training the second detection model using the second pluralities of test variables and the plurality of associated second diagnoses;
• for each of the plurality of patients:
• calculating, by applying the trained first detection model on the first plurality of test variables, a first impairment probability;
• calculating, by applying the trained second detection model on the second plurality of test variables, a second impairment probability;
• obtaining a final diagnosis made by the clinical practitioner listening to the recorded utterances made by the patient during the first and second neuropsychological tests;
• and
• associating the first impairment probability and second impairment probability with the clinical practitioner's final diagnosis;
• and
• training the final detection model using the first impairment probabilities, the second impairment probabilities and the plurality of associated final diagnoses.