Diagnosis assistance system, diagnosis assistance method, and diagnosis assistance program

Abstract

[Problem] To perform an assessment relating to incidence of disease in a more objective manner, in order to suitably assist diagnosis. [Solution] This diagnosis assistance system S is provided with a feature amount extraction unit (213), a classifier creation unit (214), and an assessment unit (215). The feature amount extraction unit (213) extracts, from brain waves, a feature amount having correlation with an episode of delirium. The classifier creation unit (214) creates, on the basis of a feature amount extracted from brain waves of a person experiencing an episode of delirium and a feature amount extracted from brain waves of a person who is not experiencing an episode of delirium, a classifier for assessing whether an episode of delirium is being experienced from a feature amount. The assessment unit (215) assesses whether a patient is experiencing an episode of delirium on the basis of the classifier and a feature amount extracted from brain waves of a patient being diagnosed.

Description

Diagnostic support system, diagnostic support method, and diagnostic support program

[0001] The present invention relates to a diagnosis support system, a diagnosis support method, and a diagnosis support program.

[0002] Conventionally, diagnosis of mental disorders or diseases accompanied by impaired consciousness has been performed by diagnosticians such as doctors. For example, the diagnostician conducts a conversation with a patient suspected of developing a disease, and makes a diagnosis based on the patient's condition and responses observed during the conversation. An example of a technology for assisting such diagnosticians in making a diagnosis is disclosed in Patent Literature 1. The technology disclosed in Patent Literature 1 analyzes electroencephalograms measured at a predetermined location on the patient's head to determine whether the patient is suffering from a mental disorder or disease accompanied by impaired consciousness. This can assist the diagnostician in making a diagnosis.

[0003] Special Publication No. 2019-500939

[0004] Conventional techniques focus on the known phenomenon that brain waves become slower when a patient is suffering from a disease. For example, the technique disclosed in the aforementioned Patent Document 1 identifies whether brain waves are becoming slower based on the ratio of the number of occurrences of low-frequency and high-frequency waves in the brain waves, and determines whether or not the patient is suffering from a disease based on this. However, slowing down is only one of the changes in brain waves that occur when a patient is suffering from a disease, and it is thought that there are other changes in brain waves that are related to the onset of a disease. Therefore, it is desirable to make a more objective determination of the onset of a disease without relying solely on the known phenomenon of slowing down.

[0005] The present invention has been made in view of the above circumstances, and an object of the present invention is to more objectively determine the onset of a disease in order to appropriately support diagnosis.

[0006] In order to solve the above problem, a diagnostic support system according to one embodiment of the present invention comprises: a feature extraction means for extracting features from electroencephalograms that are correlated with the onset of delirium or the possibility of the onset of delirium; a classifier creation means for creating a classifier based on the features extracted from the electroencephalograms of patients who have developed delirium or who are likely to develop delirium, and the features extracted from the electroencephalograms of patients who have not developed delirium, to determine whether a patient has developed delirium or is likely to develop delirium from the features; and a determination means for determining whether a patient to be diagnosed has delirium or is likely to develop delirium based on the features extracted from the electroencephalograms of the patient and the classifier.

[0007] According to the present invention, it is possible to more objectively determine the onset of a disease in order to appropriately support diagnosis.

[0008] FIG. 1 is a block diagram showing an example of the overall configuration of a diagnostic support system according to one embodiment of the present invention. FIG. 2 is a block diagram showing an example of the configuration of an electroencephalogram (EEG) measuring device according to one embodiment of the present invention. FIG. 3 is a graph showing the results of an EEG measuring device according to one embodiment of the present invention performing noise removal on an electroencephalogram using a Hampel filter. FIG. 4 is a block diagram showing an example of the configuration of a diagnostic support device according to one embodiment of the present invention. FIG. 5 is a graph showing a schematic diagram of a diagnostic support device according to one embodiment of the present invention determining optimal sensitivity and specificity in an ROC curve. FIG. 6 is a flowchart showing the flow of measurement processing performed by an electroencephalogram determination device according to one embodiment of the present invention. FIG. 7 is a flowchart showing the flow of classifier creation processing performed by a diagnostic support device according to one embodiment of the present invention. FIG. 8 is a flowchart showing the flow of diagnostic support processing performed by a diagnostic support device according to one embodiment of the present invention.

[0009] An example of an embodiment of the present invention will now be described with reference to the accompanying drawings.

[0010] [System Configuration] Fig. 1 is a block diagram showing the overall configuration of a diagnosis support system S according to this embodiment. As shown in Fig. 1, the diagnosis support system S includes an electroencephalogram (EEG) measuring device 10 and a diagnosis support device 20. Fig. 1 also shows a user who receives diagnostic support from the diagnosis support system S, and a subject whose EEG is measured by the diagnosis support system S.

[0011] The electroencephalogram measuring device 10 and the diagnostic support device 20 are communicatively connected in accordance with any communication method. This communication may be performed directly between the devices or via a network including a relay device. When communication is performed via a network, this network may be realized by, for example, a network such as the Internet or a LAN (Local Area Network), or a network combining these.

[0012] The diagnostic support system S is an example of an embodiment of the present invention and supports a user's diagnosis. As a result of extensive testing and research into diagnostic support, the inventors of the present invention discovered a correlation between mental disorders and diseases accompanied by impaired consciousness and feature quantities extracted from electroencephalograms (EEG). This led to the invention of a more objective determination of the onset of such diseases based on EEG, without relying solely on known phenomena such as slow wave activity. Accordingly, in this embodiment, as an example for illustrative purposes, it is assumed that the diagnostic support system S supports the diagnosis of delirium, which is one of the diseases accompanied by mental disorders and impaired consciousness. However, delirium is merely one example of a disease accompanied by mental disorders and impaired consciousness, and using delirium as an example is not intended to limit the scope of application of the present invention.

[0013] The EEG measurement device 10 measures fluctuations in electrical potential in the head of a subject as the subject's brain waves. The EEG measurement device 10 is configured as a headset-type electroencephalograph, equipped with a pair of electrodes (e.g., a pair of electrodes consisting of a first reference electrode and a second probe electrode) for measuring the subject's brain waves, or a larger number of electrodes, each of which electrically contacts a predetermined part of the subject. The EEG measurement device 10 measures the subject's brain waves using these electrodes, thereby generating data corresponding to the subject's brain waves (hereinafter referred to as "brain wave data"). The EEG measurement device 10 also transmits the generated brain wave data to the diagnosis support device 20.

[0014] The diagnostic support device 20 is a device that determines whether or not a subject has developed delirium based on the EEG data generated by the EEG measuring device 10. The diagnostic support device 20 is configured, for example, by an information processing device such as a personal computer or a server device. To determine whether or not a subject has developed delirium, the diagnostic support device 20 extracts features correlated with the onset of delirium from the EEG data measured by the EEG measuring device 10. The diagnostic support device 20 also creates a classifier that determines whether or not a subject has developed delirium based on the features extracted from the EEG data of a subject with delirium and the features extracted from the EEG data of a subject without delirium (including patients who are not likely to develop delirium). The diagnostic support device 20 then determines whether or not a subject has developed delirium based on the features extracted from the EEG of the subject to be diagnosed and the classifier.

[0015] The diagnostic support device 20 may be a device that determines whether or not there is a possibility of delirium onset based on the EEG data generated by the EEG measuring device 10. To determine whether or not there is a possibility of delirium onset, the diagnostic support device 20 extracts feature quantities correlated with the possibility of delirium onset from the EEG data measured by the EEG measuring device 10. Furthermore, the diagnostic support device 20 creates a classifier that determines whether or not there is a possibility of delirium onset based on the feature quantities extracted from the EEG data of patients who may be onset of delirium and the feature quantities extracted from the EEG data of individuals who have not developed delirium (including patients who are not at risk of developing delirium). Furthermore, the diagnostic support device 20 determines whether or not there is a possibility of delirium onset based on the feature quantities extracted from the EEG of the subject to be diagnosed and the classifier.

[0016] In this way, in the diagnosis support system S, the EEG measuring device 10 measures EEG data. Also, in the diagnosis support system S, the diagnosis support device 20 creates a classifier based on features extracted from the EEG data, thereby making a more objective determination regarding the onset of delirium without relying solely on known phenomena such as slow waves. Therefore, the diagnosis support system S can make a more objective determination regarding the onset of disease in order to appropriately support a diagnosis.

[0017] Furthermore, the diagnosis support system S can provide more objective disease assessments to appropriately support diagnosis, thereby resolving common problems with the prior art. For example, it can solve the problem that slow waves are only a part of the changes in EEG that occur when a patient is affected by a disease, and other changes in EEG that are related to the onset of a disease are not taken into account in the assessment.

[0018] Next, the configurations, functions, etc. of the EEG measuring device 10 and the diagnostic support device 20 for realizing such processing will be described in more detail. For clarity of explanation, a user (e.g., a medical professional such as a doctor) who receives diagnostic support from the diagnostic support system S will be referred to as a "diagnostician." Furthermore, a subject whose EEG data is measured so that the diagnostic support system S can create a classifier (e.g., a collaborator in creating the classifier) ​​will be referred to as a "subject." Furthermore, a subject whose delirium is determined by the classifier created by the diagnostic support system S (e.g., a person suspected of developing a mental disorder or a disease accompanied by impaired consciousness) will be referred to as a "patient."

[0019] [Configuration of EEG Measuring Device] The configuration of the EEG measuring device 10 will be described with reference to Fig. 2. Fig. 2 is a block diagram showing an example of the configuration of the EEG measuring device 10. As shown in Fig. 2, the EEG measuring device 10 includes a CPU (Central Processing Unit) 11, a ROM (Read Only Memory) 12, a RAM (Random Access Memory) 13, a communication unit 14, a storage unit 15, an input unit 16, an output unit 17, and a measurement unit 18. These units are connected by signal lines and transmit and receive signals to and from each other.

[0020] The CPU 11 executes various processes (for example, measurement processes described below) according to programs recorded in the ROM 12 or programs loaded from the storage unit 15 to the RAM 13. The RAM 13 also stores data and the like necessary for the CPU 11 to execute various processes.

[0021] The communication unit 14 controls communication between the CPU 11 and other devices (e.g., the diagnosis support device 20). The storage unit 15 is configured with a semiconductor memory such as a DRAM (Dynamic Random Access Memory) and stores various data.

[0022] The input unit 16 is made up of various buttons etc. and inputs various information in response to user operations. The output unit 17 is made up of a display, a speaker etc. and outputs images and sounds.

[0023] The measurement unit 18 measures fluctuations in the electrical potential of the head of the person being measured (here, a test subject or patient) as the person's electroencephalogram (EEG). In this embodiment, as an example of a measurement method, it is assumed that the measurement unit 18 measures monopolar EEGs using a reference electrode derivation method. In this case, one end of a pair of electrodes provided by the measurement unit 18 is brought into contact with a position where the electrical potential is close to zero (e.g., the person's earlobe) to serve as the reference electrode. The other end is brought into contact with a predetermined position on the person's head (e.g., a position corresponding to Fp1 in the left prefrontal cortex as defined by the International 10-20 System) to serve as the probe electrode. The measurement unit 18 then measures fluctuations in the electrical potential difference between the reference electrode and the probe electrode over time at a predetermined sampling frequency (e.g., 512 Hz) as EEGs in a predetermined region of the person's brain.

[0024] As described above, in this embodiment, it is possible to determine whether or not a subject has developed delirium simply by measuring the EEG in a single region of the subject's brain using a pair of electrodes. This is because the diagnostic support device 20, described below, can make this determination using only EEG data from a single location, without performing processing such as comparing EEG data from multiple locations. Because only a pair of electrodes is required, the EEG measuring device 10 in this embodiment can have a simple, headset-like shape. Furthermore, by performing appropriate noise processing, the EEG measuring device 10 can also have an even simpler shape, such as a device that can measure EEG data simply by placing electrodes on the forehead.

[0025] As described above, in this embodiment, the EEG measuring device 10 can be configured as a headset. In this case, when the subject wears the EEG measuring device 10, the pair of electrodes provided in the measuring unit 18 are positioned at positions suitable for measurement (e.g., a position contacting the earlobe and a position contacting the area corresponding to Fp1). In this regard, using a typical EEG device that covers the subject's head with an electrode net can cause a sense of pressure on the subject. In addition, it takes time to adjust the positions of the multiple electrodes so that each is positioned appropriately, which places a burden on the subject. Furthermore, due to these factors, when using a typical EEG device, noise due to tension is generated in the EEG. In contrast, a headset configuration can prevent such a sense of pressure and enable measurement while suppressing tension in the subject. This is also true for a configuration that allows EEG measurement by simply placing electrodes on the forehead. Therefore, the EEG measuring device 10 can suppress noise due to tension in the subject and perform accurate measurement.

[0026] In the EEG measuring device 10, these units cooperate to perform a "measurement process." Here, the measurement process is a series of steps that measure the EEG of the subject and perform predetermined pre-processing on the measured EEG.

[0027] 2, when the measurement process is executed, a measurement control unit 111, a preprocessing unit 112, and an electroencephalogram data transmission unit 113 function in the CPU 11. Data required to realize the process is transmitted and received between these functional blocks at appropriate times, including cases not specifically mentioned below.

[0028] The measurement control unit 111 controls the measurement of brain waves by the measurement unit 18 based on instructions from the subject or diagnostician received by the input unit 16 (or instructions from the subject or diagnostician received via the communication unit 14). For example, based on these instructions, the measurement control unit 111 controls the timing of the start and end of measurement by the measurement unit 18, and controls the sampling period in the measurement, etc. Then, the measurement control unit 111 outputs the brain waves of the subject obtained by measurement by the measurement unit 18 to the preprocessing unit 112.

[0029] The measurement of EEG signals by the measurement control unit 111 may be performed at any time, for example, before or after the subject undergoes surgery or enters an ICU (Intensive Care Unit), for example, every other day. The length of each measurement is also arbitrary, but measurements may be performed for approximately three minutes each under different circumstances, such as when the subject is resting with their eyes closed or in other situations. In this case, the subject's posture is, for example, a seated position, but if it is difficult for the subject to get out of bed, other positions such as supine may also be used. In this way, by completing the measurement in a short time, for example, three minutes, it is possible to perform the measurement without any problems even if the subject is restless and long-term measurement is difficult.

[0030] The preprocessing unit 112 generates EEG data by performing preprocessing, such as removing noise components, on the EEG input from the measurement control unit 111. To this end, for example, the preprocessing unit 112 uses a predetermined filter to detect outliers from the EEG input from the measurement control unit 111 and correct the detected outliers. In this embodiment, a Hampel filter is used as the predetermined filter. FIG. 3 is a graph showing the results of noise removal using a Hampel filter on the EEG input from the measurement control unit 111. In each of FIGS. 3A and 3B, the horizontal axis represents time [S], and the vertical axis represents the value [μV] of the potential difference between the reference electrode and the exploration electrode measured as the EEG by the measurement unit 18 (i.e., the measured value).

[0031] When using a Hampel filter, the preprocessing unit 112 calculates the local median of the measured values ​​over a five-second window, for example. The preprocessing unit 112 also detects, as outliers, any measured values ​​exceeding twice the standard deviation of the calculated local median. The preprocessing unit 112 then replaces the detected outlier with the local median over the five-second window. The preprocessing unit 112 then repeats this process while moving the window, thereby removing any outliers (spikes) contained in the measured values ​​as noise components. The results of this noise removal are shown in FIG. 3A . Furthermore, the results of noise removal when similar processing is performed by detecting, as outliers, any measured values ​​exceeding three times the standard deviation of the calculated local median, rather than two times, the standard deviation. In either case, the outliers (spikes) contained in the measured values, which are noise components, are almost completely removed. In this embodiment, by using a Hampel filter as the predetermined filter, noise components can be appropriately removed. Therefore, as described above, the EEG measuring device 10 can be made into a simple headset-type shape or an even simpler shape, such as a device that can measure EEGs by simply placing electrodes on the forehead. Furthermore, a Hampel filter has the characteristic of not easily smoothing the data (measured values, in this case) required for removing noise components. Therefore, compared to using a filter such as a median filter, which tends to smooth the required data (measured values, in this case), the use of a Hampel filter makes it possible to obtain data corresponding to actual EEGs with high accuracy.

[0032] Next, for example, the pre-processing unit 112 uses a band-pass filter to pass only a predetermined frequency band (for example, 1 to 30 Hz) from the electroencephalograms whose outliers have been corrected by the Hampel filter.

[0033] It is generally known that brain waves slow down during sleep, regardless of whether a person is experiencing delirium. Therefore, if brain waves in the delta band (0.5 Hz to 4 Hz), which indicate slow waves, are used in subsequent processing, it may be difficult to distinguish whether a person is experiencing delirium or simply sleeping. Therefore, the preprocessing unit 112 may set the lower limit of the frequency band to be passed to, for example, 5 Hz, and not use brain waves below 5 Hz in subsequent processing. This makes it possible to accurately determine whether a person is experiencing delirium, even from brain waves measured during sleep.

[0034] Furthermore, it is generally known that when EEG measurements are performed on a subject in a position other than the supine position, noise such as myoelectric noise occurs in a frequency band of, for example, 20 Hz or higher. Therefore, the preprocessing unit 112 may set the upper limit of the frequency band to pass through to, for example, 20 Hz, and may not use EEG signals exceeding 20 Hz for further processing. This makes it possible to accurately determine whether or not a subject has developed delirium, without being affected by noise such as myoelectric noise, even when EEG measurements are performed in a position where noise is likely to occur, such as a sitting position.

[0035] In this way, in this embodiment, the onset of delirium can be determined more objectively without relying solely on known phenomena such as slow-wave changes. Therefore, for example, it is possible to avoid using frequency bands that indicate slow-wave changes in processing. However, these upper and lower limit values ​​are merely examples. For example, if it is clear that the subject is awake and not asleep at the time of measurement, the lower limit does not necessarily have to be 5 Hz. Similarly, the upper limit does not necessarily have to be 20 Hz as long as noise removal can be performed appropriately.

[0036] In this way, the preprocessing unit 112 generates electroencephalogram data from the measured electroencephalograms by performing preprocessing such as removing noise components using a Hampel filter or a bandpass filter, and then outputs the generated electroencephalogram data to the electroencephalogram data transmitting unit 113.

[0037] The electroencephalogram data transmitting unit 113 transmits the electroencephalogram data input from the preprocessing unit 112 to the diagnosis support device 20. The transmission may be performed in real time each time electroencephalogram data is generated by the preprocessing unit 112, or the generated electroencephalogram data may be stored in the storage unit 15, and the electroencephalogram data stored in the storage unit 15 may be transmitted all at once at the end of measurement or other timing.

[0038] [Configuration of the diagnosis support device] The configuration of the diagnosis support device 20 will be described with reference to Fig. 4. Fig. 4 is a block diagram showing an example of the configuration of the diagnosis support device 20. As shown in Fig. 4, the diagnosis support device 20 includes a CPU 21, a ROM 22, a RAM 23, a communication unit 24, a storage unit 25, an input unit 26, an output unit 27, and a drive 28. These units are connected by signal lines and send and receive signals to and from each other.

[0039] The CPU 21 executes various processes (for example, a classifier creation process and a diagnostic support process, which will be described later) according to a program recorded in the ROM 22 or a program loaded from the storage unit 25 to the RAM 23. The RAM 23 also stores data and the like necessary for the CPU 21 to execute various processes, as appropriate.

[0040] The communication unit 24 controls communication so that the CPU 21 can communicate with other devices (for example, the electroencephalogram (EEG) measurement device 10). The storage unit 25 is configured with a semiconductor memory such as a DRAM (Dynamic Random Access Memory) and stores various data.

[0041] The input unit 26 is composed of various buttons and a touch panel, or external input devices such as a mouse and a keyboard, and inputs various information in response to user instructions. The output unit 27 is composed of a display, a speaker, etc., and outputs images and sounds.

[0042] Removable media (not shown) such as a magnetic disk, optical disk, magneto-optical disk, or semiconductor memory are appropriately loaded into the drive 28. The programs read from the removable media by the drive 28 are installed in the storage unit 25 as needed.

[0043] In the diagnostic support device 20, these units cooperate to perform a "classifier creation process" and a "diagnosis support process." Here, the classifier creation process is a series of processes in which the diagnostic support device 20 creates a classifier that determines whether or not a subject has developed delirium, based on the EEG data of the subject generated by the EEG-measuring device 10. Also, the diagnostic support process is a series of processes in which the diagnostic support device 20 determines whether or not a patient to be diagnosed has developed delirium, based on the EEG data of the patient generated by the EEG-measuring device 10 and the classifier. That is, in this embodiment, the diagnostic support device 20 first creates a classifier, and then uses the created classifier to more objectively determine the onset of delirium without relying solely on known phenomena such as slow wave transition.

[0044] When the classifier creation process and the diagnostic support process are executed, an electroencephalogram data acquisition unit 211, a diagnostic result acquisition unit 212, a feature extraction unit 213, a classifier creation unit 214, and a determination unit 215 function in the CPU 21, as shown in Fig. 4. An electroencephalogram data storage unit 251, a feature storage unit 252, and a classifier storage unit 253 are provided in one area of ​​the storage unit 25. Data required to realize the processes is transmitted and received between these functional blocks at appropriate times, including cases not specifically mentioned below.

[0045] The electroencephalogram data acquiring section 211 acquires the electroencephalogram data by receiving it transmitted from the electroencephalogram measuring device 10. Then, the electroencephalogram data acquiring section 211 stores the acquired electroencephalogram data in the electroencephalogram data storage section 251. In other words, the electroencephalogram data storage section 251 functions as a storage section that stores the electroencephalogram data.

[0046] The EEG measuring device 10 may be provided with a drive similar to the drive 28 so that the EEG measuring device 10 stores the EEG data in removable media. The EEG data acquiring section 211 may acquire the EEG data from this removable media via the drive 28, rather than acquiring the data through communication.

[0047] The diagnostic result acquisition unit 212 acquires a diagnostic result as to whether or not each subject corresponding to the EEG data has developed delirium. This diagnostic result is not a diagnostic result based on assistance from the diagnostic support process, but a diagnostic result based on a normal method such as a doctor's interview. That is, in this embodiment, the diagnosis of the patient is performed based on assistance from the diagnostic support process, while the diagnosis of the subject is performed based on a normal method such as a doctor's interview. The normal method is, for example, a method based on an index such as CAM-ICU (confusion assessment method for the ICU) or ICDSC (intensive care delirium screening checklist). Here, in order for the classifier creation unit 214 (described later) to create a highly accurate classifier, it is desirable that the diagnostic result acquired by the diagnostic result acquisition unit 212 be a highly reliable diagnostic result. Therefore, the diagnostic results acquired by the diagnostic result acquisition unit 212 should be more reliable, such as diagnostic results based on a comprehensive judgment by multiple doctors, diagnostic results by an experienced doctor, or diagnostic results that take into account the results of tests such as blood tests and imaging tests.

[0048] The diagnostic result acquisition unit 212 acquires the diagnostic results for each subject using such conventional methods, for example, based on input operations from the diagnostician accepted by the input unit 26 (or input operations from the diagnostician received via the communication unit 14).

[0049] Furthermore, the diagnostic result acquisition unit 212 stores the acquired diagnostic results for each subject in the electroencephalogram data storage unit 251. In this case, the diagnostic results for each subject and the electroencephalogram data corresponding to that subject are linked and stored in the electroencephalogram data storage unit 251. This makes it possible to distinguish the electroencephalogram data for each subject stored in the electroencephalogram data storage unit 251 into electroencephalogram data for subjects who have developed delirium (i.e., subjects who are positive for delirium) and electroencephalogram data for subjects who have not developed delirium (i.e., subjects who are negative for delirium).

[0050] The feature extraction unit 213 extracts features correlated with the onset of delirium from the EEG data of the subject and the patient stored in the EEG data storage unit 251. In this case, for example, the feature extraction unit 213 divides the EEG data into predetermined time units (e.g., 20 seconds). Furthermore, for example, the feature extraction unit 213 extracts frequency components of each of a plurality of frequencies from the divided EEG data. To this end, the feature extraction unit 213 performs a Fourier transform (e.g., a fast Fourier transform (FFT) using a 512-point Hamming window and 50% overlap processing) on ​​the divided EEG data and averages the results to generate a power spectrum indicating the power values ​​of each of a plurality of frequencies as the frequency components in the EEG data.

[0051] Furthermore, taking into consideration individual differences in the amplitude of the electroencephalogram data, the feature extraction unit 213 normalizes the power spectrum. For example, the feature extraction unit 213 calculates the proportion of the power value occupied by a corresponding frequency in the electroencephalogram data, and normalizes the power value so that the minimum value is 0 and the maximum value is 1.

[0052] Next, the feature extraction unit 213 selects which frequency power value in the normalized power spectrum to extract as a feature. In this case, it is desirable to extract as a feature the power value of a frequency that is more highly correlated with the onset of delirium. Therefore, the feature extraction unit 213 performs a test to identify frequencies that have a significant difference in relation to whether or not a subject has delirium, using the power values ​​of each of multiple frequencies in the power spectrum. For example, the feature extraction unit 213 performs a Wilcoxon rank-sum test on the power values ​​of the EEG data of subjects with delirium and the power values ​​extracted from the EEG data of subjects without delirium, based on a predetermined significance level (e.g., significance level α = 0.01), to identify only frequencies that have a predetermined significant difference. Then, the power values ​​(i.e., frequency components) of the frequencies that have the predetermined significant difference are extracted as features correlated with the onset of delirium.

[0053] In this way, the feature extraction unit 213 can extract features that are correlated with the onset of delirium from the subject's EEG data or the patient's EEG data by identifying frequencies that have significant differences regarding whether or not delirium has occurred.

[0054] In this embodiment, instead of extracting feature quantities correlated with the onset of delirium, feature quantities correlated with the possibility of onset of delirium may be extracted, and the following processes, such as creating a classifier and making a determination using the classifier, may be performed. In this case, the diagnosis result acquisition unit 212 acquires a diagnosis result indicating whether or not there is a possibility of onset of delirium for each subject corresponding to the EEG data. As a result, the EEG data for each subject stored in the EEG data storage unit 251 can be distinguished into EEG data of patients who are likely to develop delirium and EEG data of patients who are not likely to develop delirium. Then, the feature extraction unit 213 extracts feature quantities correlated with the possibility of onset of delirium from the EEG data of the subject and the EEG data of the patient stored in the EEG data storage unit 251. In this way, it becomes possible to determine whether or not there is a possibility of onset of delirium for a subject based on feature quantities correlated with the possibility of onset of delirium.

[0055] Then, the feature extraction unit 213 stores the extracted feature amounts of individuals with delirium and the extracted feature amounts of individuals without delirium in the feature storage unit 252. In other words, the feature storage unit 252 functions as a storage unit that stores feature amounts extracted from electroencephalogram data.

[0056] The classifier creation unit 214 creates a classifier for determining whether or not a subject has delirium based on the features extracted from the EEG data of subjects with delirium and the features extracted from the EEG data of subjects without delirium, which are stored in the feature extraction unit 213. The classifier may be, for example, a classifier that realizes binary classification (here, classification of whether or not a subject has delirium), and the creation method is not particularly limited. For example, the classifier creation unit 214 creates the classifier using logistic regression analysis. In this case, the classifier creation unit 214 optimizes explanatory variables using a variable addition / decrement method in the logistic regression equation shown in the following equation (1). Here, the explanatory variables are frequencies corresponding to the features extracted by the feature storage unit 252, and it optimizes which frequencies to select as explanatory variables.

[0057] In equation (1), p is the positive probability, and β ｉ is the partial regression coefficient, and x ｉ is an explanatory variable.

[0058] Furthermore, for example, the classifier creation unit 214 creates a receiver operating characteristic (ROC) curve as an index of the predictive ability of the classifier, and calculates the area under the curve (AUC). ｃ If p > p, delirium was considered non-occurring (i.e., negative), and the cutoff value p ｃ If p is less than or equal to the threshold, delirium is diagnosed (i.e., positive), and based on these results, the sensitivity (i.e., the probability that a normal person can be determined to be normal) and specificity (i.e., the probability that an abnormal person can be determined to be abnormal) are determined.

[0059] Furthermore, for example, the classifier creation unit 214 may perform bootstrapping, which is a method of estimating a population from a sample population, using the explanatory variable x ｉ For example, the classifier creation unit 214 derives a confidence interval by performing resampling a predetermined number of times (for example, 2000 times) on the partial regression coefficient β ｉ, and derive the 95% confidence interval of AUC. FIG. 5 is a graph that schematically illustrates how to determine the optimal sensitivity and specificity in an ROC curve. In this case, for example, the classifier creation unit 214 determines the optimal sensitivity and specificity in the ROC curve shown in FIG. 5 using the Index of Union method as shown in the following formula (2).

[0060]

[0061] Details of such bootstrapping using the Index of Union method are disclosed in, for example, Non-Patent Document 1 below.

[0062] <Non-patent Document 1> Ilker Unal, “Defining an Optimal Cut-Point Value in ROC Analysis: An Alternative Approach”, [online], Computational and Mathematical Methods in Medicine, Vol. 2017, 2017, [Retrieved September 29, 2021], Internet <URL: https: / / www.hindawi.com / journals / cmmm / 2017 / 3762651 / >

[0063] By performing this series of processes, the classifier creation unit 214 performs optimization for the logistic regression equation shown in the above equation (1) and calculates the explanatory variable x ｉ and the partial regression coefficient β ｉ The classifier creation unit 214 stores the logistic regression equation optimized in this manner in the classifier storage unit 253 as a classifier that determines whether or not delirium has developed from the feature amount. That is, the classifier storage unit 253 functions as a storage unit that stores the classifier generated by the classifier creation unit 214 performing this series of processes.

[0064] In this way, when optimizing the logistic regression equation based on more objective indicators, such as features extracted from EEG data, the explanatory variable x ｉAs the feature quantity, both frequencies corresponding to feature quantities whose values ​​increase with the onset of delirium and frequencies corresponding to feature quantities whose values ​​decrease with the onset of delirium may be selected. For example, both frequencies whose power value, which is a feature quantity, decreases with a temporary decline in cognitive function due to delirium and frequencies whose power value, which is a feature quantity, increases with stress due to anxiety or pain caused by delirium may be selected. That is, the feature quantity in this embodiment may include both a first feature quantity whose value increases with the onset of delirium and a second feature quantity whose value decreases with the onset of delirium. As described above, in this embodiment, the use of objective indices in creating the classifier makes it possible to extract feature quantities based on multiple perspectives, thereby enabling the creation of a classifier that makes more accurate judgments.

[0065] The determination unit 215 determines whether or not the patient has developed delirium based on the features extracted from the EEG of the patient to be diagnosed stored in the feature storage unit 252 and the classifier stored in the classifier storage unit 253. For example, the determination unit 215 acquires the features extracted from the EEG of the patient to be diagnosed based on a selection operation of the patient to be diagnosed received by the input unit 26 (or a selection operation of the patient to be diagnosed from the diagnosing expert received via the communication unit 24). The determination unit 215 also acquires the classifier by reading it from the feature storage unit 252. The determination unit 215 then inputs the acquired features into corresponding explanatory variables in the acquired classifier. The determination unit 215 then determines whether or not the patient has developed delirium based on the classification result of the classifier (i.e., the value of the objective variable calculated in the classifier).

[0066] The determination unit 215 then presents the determination result to the diagnostician. This presentation may be, for example, a display on a display included in the output unit 27, an audio output from a speaker included in the output unit 27, printing on a paper medium from a printing device via the communication unit 24, or transmission to another device (not shown) used by the diagnostician via the communication unit 24.

[0067] In this way, the diagnostic support device 20 creates a classifier based on features extracted from EEG data, thereby making a more objective determination of the onset of delirium without relying solely on known phenomena such as slow waves. By referencing the determination results based on these objective indicators, the diagnostician can make a more appropriate diagnosis, rather than relying solely on his or her own subjective judgment. Therefore, according to this embodiment, a more objective determination of the onset of a disease can be made to appropriately support the diagnosis.

[0068] [Measurement Processing] Next, the flow of the measurement processing executed by the EEG measuring device 10 will be described with reference to Fig. 6. Fig. 6 is a flowchart illustrating the flow of the measurement processing executed by the EEG measuring device 10. The measurement processing is executed in response to an instruction operation to start measurement from the subject or the diagnostician.

[0069] In step S11, the measurement control unit 111 controls the measurement of the subject's brain waves by the measurement unit 18, thereby starting measurement of the brain waves by the measurement unit 18. Then, the measurement control unit 111 outputs the subject's brain waves obtained by measurement by the measurement unit 18 to the preprocessing unit 112.

[0070] In step S12, the preprocessing unit 112 generates electroencephalogram data by performing preprocessing on the electroencephalograms input from the measurement control unit 111. Then, the preprocessing unit 112 outputs the generated electroencephalogram data to the electroencephalogram data transmitting unit 113.

[0071] In step S13, the electroencephalogram data transmitting unit 113 transmits the electroencephalogram data input from the preprocessing unit 112 to the diagnosis support device 20. Note that, in this flowchart, it is assumed that the electroencephalogram data is transmitted in real time each time it is generated by the preprocessing unit 112, but as described above, it is also possible to store the generated electroencephalogram data in the storage unit 15 and transmit all of the electroencephalogram data stored in the storage unit 15 at once, for example, when the measurement is completed.

[0072] In step S14, the measurement control unit 111 determines whether or not to end the measurement of electroencephalograms by the measurement unit 18. For example, the measurement control unit 111 determines to end the measurement of electroencephalograms when a predetermined time has elapsed since the start of measurement, or when the subject or the diagnostician issues an instruction to end the measurement. If the measurement of electroencephalograms is to be ended, step S14 is determined as Yes, and the process ends. On the other hand, if the measurement of electroencephalograms is not to be ended, step S14 is determined as No, and the process is repeated again from step S11.

[0073] Through the measurement process described above, the EEG measurement device 10 can measure the EEG of the subject and transmit EEG data generated based on the measured EEG to the diagnosis support device 20.

[0074] [Classifier Creation Process] Next, the flow of the classifier creation process executed by the diagnosis support device 20 will be described with reference to Fig. 7. Fig. 7 is a flowchart illustrating the flow of the classifier creation process executed by the diagnosis support device 20. The classifier creation process is executed in response to an instruction operation to start classifier creation from a diagnostician or an administrator of the diagnosis support system S. Note that, as a premise of the process, it is assumed that the electroencephalogram data of the subject generated by the measurement process has been received by the electroencephalogram data acquisition unit 211 and stored in the electroencephalogram data storage unit 251.

[0075] In step S21, the diagnostic result acquisition unit 212 acquires the diagnostic result of the subject based on an input operation from the diagnostician.

[0076] In step S22, the feature extraction section 213 extracts features from the electroencephalogram data of the subject.

[0077] In step S23, the classifier creation unit 214 starts creating a classifier that determines whether or not a person has developed delirium based on the features extracted from the EEG data of individuals who have developed delirium and the features extracted from the EEG data of individuals who have not developed delirium.

[0078] In step S24, the classifier creation unit 214 performs a series of processes such as bootstrapping as described above to optimize the classifier.

[0079] In step S25, the classifier creation unit 214 determines whether or not the optimization of the classifier has been completed. For example, the optimization of the classifier is completed when the number of optimization attempts reaches a predetermined number. If the optimization of the classifier has been completed, the determination in step S25 is Yes, and the process proceeds to step S26. On the other hand, if the optimization of the classifier has not been completed, the determination in step S25 is No, and the process is repeated again from step S24.

[0080] In step S26, the classifier creation unit 214 stores the optimized classifier in the classifier storage unit 253. This completes the process.

[0081] By the classifier creation process described above, the diagnosis support device 20 can create a classifier that the determination unit 215 uses to determine whether or not delirium has developed from the feature amount.

[0082] [Diagnostic Support Processing] Next, the flow of the diagnostic support processing executed by the diagnostic support device 20 will be described with reference to Fig. 8. Fig. 8 is a flowchart illustrating the flow of the diagnostic support processing executed by the diagnostic support device 20. The diagnostic support processing is executed in response to an instruction operation from the diagnostician to start diagnostic support. Note that, as a premise of the processing, it is assumed that the patient's electroencephalogram data generated by the measurement processing is received by the electroencephalogram data acquisition unit 211 and stored in the electroencephalogram data storage unit 251. It is also assumed that the classifier generated by the classifier generation processing is stored in the classifier storage unit 253.

[0083] In step S31, the feature extraction section 213 extracts feature amounts from the electroencephalogram data of the patient to be diagnosed.

[0084] In step S32, the determination unit 215 determines whether or not the patient to be diagnosed has developed delirium based on the feature amounts extracted from the electroencephalogram data of the patient and the classifier.

[0085] In step S33, the determination unit 215 presents the result of the determination made in step S32 to the person making the diagnosis.

[0086] Through the diagnostic support process described above, the diagnostic support device 20 creates a classifier to more objectively determine the onset of delirium without relying solely on known phenomena such as slow wave activity. By referencing this more objective determination result, the diagnostician can make a more appropriate diagnosis, rather than relying solely on their own judgment. Therefore, according to this embodiment, a more objective determination of disease onset can be made to appropriately support the diagnosis.

[0087] [Modifications] Although the embodiments of the present invention have been described above, these embodiments are merely illustrative and do not limit the technical scope of the present invention. The present invention can take on various other embodiments without departing from the spirit of the present invention, and various modifications such as omissions and substitutions can be made. In such cases, these embodiments and their modifications are included in the scope and spirit of the invention described in the present specification, etc., and are also included in the scope of the invention and its equivalents described in the claims. As an example, the above-described embodiments of the present invention may be modified as exemplified below.

[0088] The device configuration of the diagnostic support system S in the above-described embodiment is merely an example and can be modified as appropriate. For example, in the above-described embodiment, the preprocessing unit 112 in the EEG measuring device 10 performs preprocessing such as noise removal. This is not a limitation, but for example, a functional block corresponding to the preprocessing unit 112 may be operated in the diagnostic support device 20, and this functional block may perform preprocessing such as noise removal. In addition, in the above-described embodiment, the EEG measuring device 10 and the diagnostic support device 20 are realized as separate devices, but this is not a limitation, but for example, the EEG measuring device 10 and the diagnostic support device 20 may be realized as an integrated device.

[0089] In the above-described embodiment, the EEG measuring device 10 and the diagnostic support device 20 are each implemented by a single computer. However, the present invention is not limited to this, and each of the EEG measuring device 10 and the diagnostic support device 20 may be implemented by multiple computers by using technology such as cloud computing.

[0090] In addition, in the above-described embodiment, the diagnostic support system S is configured with both the EEG measuring device 10 and the diagnostic support device 20. However, this is not limiting, and the diagnostic support system S may be configured with only the diagnostic support device 20. In this case, the EEG measuring device 10 is installed, for example, in a medical facility such as a hospital or the subject's home, which is different from the facility where the diagnostic support system S is installed. Then, at these medical facilities, homes, etc., the EEG measuring device 10 measures brain waves. Then, the electroencephalogram data of the brain waves obtained by this measurement is input to the diagnostic support device 20 constituting the diagnostic support system S via a network, a recording medium, etc. Then, the diagnostic support device 20 performs processes such as creating the classifier and making judgments described above based on this electroencephalogram data. As a result, even if the diagnostic support system S is configured with only the diagnostic support device 20, it is possible to appropriately support a diagnosis, as in the above-described embodiment.

[0091] Alternatively, this may be further modified so that, for example, only the functions of some of the functional blocks in the diagnostic support device 20 are realized by the diagnostic support device 20 constituting the diagnostic support system S, and the functions of the other functional blocks in the diagnostic support device 20 are realized by another device constituting an external system different from the functional diagnostic support system S. For example, the diagnostic support device 20 constituting the diagnostic support system S may realize only the functions of the determination unit 215 and the classifier storage unit 253. In this case, the other functional blocks realized in the external system create the classifier. The classifier storage unit 253 realized by the diagnostic support system S then acquires the created classifier via a network, a recording medium, or the like. The classifier storage unit 253 then stores the classifier. In other words, in this modification, the diagnostic support system S is provided with a classifier necessary to achieve the desired diagnosis in advance. The determination unit 215 realized by the diagnostic support system S also acquires feature quantities extracted from the electroencephalogram of the patient to be diagnosed from the external system. The determination unit 215 then makes a determination by inputting the acquired feature quantities into corresponding explanatory variables in a pre-installed classifier. As a result, even if only the determination unit 215 and the classifier storage unit 253 of the diagnosis support device 20 are realized in the diagnosis support system S, it is possible to appropriately support a diagnosis, as in the above-described embodiment. These modified examples are suitable, for example, when the operator operating the diagnosis support system S is an operator that mainly provides diagnostic support on behalf of others.

[0092] Furthermore, in the above-described embodiment, the EEG measuring device 10 measures the brain waves at Fp1 using a pair of electrodes. However, the present invention is not limited to this, and the EEG measuring device 10 may measure brain waves at a location other than Fp1. Alternatively, the EEG measuring device 10 may be provided with a larger number of electrodes and measure brain waves at a plurality of locations. Then, the above-described series of processes may be performed based on each of the brain waves at a plurality of locations.

[0093] In the above-described embodiment, the classifier is created using a logistic regression equation. However, the present invention is not limited to this, and the classifier may be created using other methods such as a simple perceptron or a linear support vector machine.

[0094] In addition, in the above-described embodiment, a classifier is created based on features extracted from EEG data, and a determination is made as to whether or not a subject has developed delirium based on objective indicators such as EEG data. However, the present invention is not limited to this, and it may be possible to determine whether or not a subject has developed a mental disorder or a disease accompanied by a disturbance of consciousness other than delirium.

[0095] [Configuration Example] As described above, the diagnosis support system S according to this embodiment includes a feature extraction unit 213, a classifier creation unit 214, and a determination unit 215. The feature extraction unit 213 extracts features correlated with the onset of delirium or the possibility of onset of delirium from EEGs. The classifier creation unit 214 creates a classifier for determining whether a patient has delirium or the possibility of onset of delirium based on the features extracted from the EEGs of patients with delirium or the possibility of onset of delirium and the features extracted from the EEGs of patients without delirium. The determination unit 215 determines whether a patient has delirium or the possibility of onset of delirium based on the features extracted from the EEGs of a patient to be diagnosed and the classifier. In this way, the diagnosis support system S creates a classifier based on the features extracted from the EEGs, thereby making a more objective determination regarding the onset of delirium without relying solely on known phenomena such as slow wave transition. Therefore, the diagnosis support system S can make a more objective judgment regarding the onset of a disease in order to appropriately support a diagnosis.

[0096] The feature extraction unit 213 extracts multiple frequency components from the electroencephalograms of a predetermined region of the brain and extracts frequency components of predetermined frequencies as features. This makes it possible to extract frequency components suitable for creating a classifier, rather than all frequency components extracted from the electroencephalograms, when creating a classifier or making a judgment, thereby making it possible to more accurately judge whether a subject has developed delirium or whether there is a possibility of developing delirium.

[0097] The classifier creation unit 214 selects some of the feature quantities in the process of creating a classifier, and creates a classifier based on the selected part of the feature quantities. This allows more appropriate feature quantities to be selected when creating a classifier, making it possible to create a classifier that can more accurately determine whether a person has developed delirium or whether there is a possibility of developing delirium.

[0098] The classifier creation unit 214 creates a classifier by performing multivariate analysis using the feature values ​​as explanatory variables and whether or not delirium has developed or is likely to develop as the objective variable. This makes it possible to determine the onset of a disease using multivariate analysis based on objective indicators rather than multiple variables extracted from the electroencephalogram.

[0099] The feature may include both a first feature whose value increases with the onset of delirium or the possibility of onset of delirium, and a second feature whose value decreases with the onset of delirium or the possibility of onset of delirium. This makes it possible to extract feature based on multiple perspectives and to create a classifier that makes more accurate judgments.

[0100] The electroencephalogram is measured by a pair of electrodes consisting of a first electrode and a second electrode. This makes it possible to easily measure the electroencephalogram with a simple structure, for example, a pair of electrodes consisting of a reference electrode serving as the first electrode and a detection electrode serving as the second electrode.

[0101] [Realization of Functions by Hardware or Software] The function of executing the series of processes according to the above-described embodiment can be realized by hardware, software, or a combination of these. In other words, it is sufficient that the function of executing the series of processes described above is realized in any of the diagnosis support systems S, and there are no particular limitations on how this function is realized.

[0102] For example, when the function of executing the above-mentioned series of processes is realized by a processor that executes arithmetic processing, the processor that executes this arithmetic processing includes processors that are configured by various processing devices alone, such as a single processor, a multiprocessor, and a multicore processor, as well as processors that combine these various processing devices with processing circuits such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field-Programmable Gate Array).

[0103] Furthermore, for example, when the function of executing the above-described series of processes is realized by software, the program constituting the software is installed on a computer via a network or a recording medium. In this case, the computer may be a computer incorporating dedicated hardware, or may be a general-purpose computer (e.g., a general electronic device such as a general-purpose personal computer) that can execute predetermined functions by installing a program. Furthermore, the steps of writing the program may include only processes that are performed chronologically according to the order, but may also include processes that are performed in parallel or individually. Furthermore, the steps of writing the program may be performed in any order within the scope of the present invention.

[0104] A recording medium on which such a program is recorded may be provided to a user by being distributed separately from the computer main unit, or may be provided to a user in a state where it is pre-installed in the computer main unit. In this case, the storage medium distributed separately from the computer main unit is composed of a magnetic disk (including a floppy disk), an optical disk, a magneto-optical disk, or the like. Optical disks include, for example, CD-ROMs (Compact Disc-Read Only Memory), DVDs (Digital Versatile Discs), and Blu-ray (registered trademark) Discs. Magneto-optical disks include, for example, MDs (Mini Discs). These storage media are installed in, for example, drive 28 in FIG. 4 and incorporated into the computer main unit. The recording medium provided to the user in a state where it is pre-installed in the computer main body is composed of, for example, the ROM 12 in FIG. 2 , the ROM 22 in FIG. 4 , the memory unit 15 in FIG. 2 , or an SSD (Solid State Drive) or hard disk included in the memory unit 25 in FIG. 4 , in which the program is recorded.

[0105] 10 EEG measuring device, 20 Diagnostic support device, 11, 21 CPU, 12, 22 ROM, 13, 23 RAM, 14, 24 Communication unit, 15, 25 Storage unit, 16, 26 Input unit, 17, 27 Output unit, 18 Measurement unit, 28 Drive, 111 Measurement control unit, 112 Preprocessing unit, 113 EEG data transmission unit, 211 EEG data acquisition unit, 212 Diagnostic result acquisition unit, 213 Feature extraction unit, 214 Classifier creation unit, 215 Determination unit, 251 EEG data storage unit, 252 Feature storage unit, 253 Classifier storage unit, S Diagnostic support system

Claims

1. a feature extraction means for extracting, from the electroencephalogram, a feature correlated with the onset of delirium or the possibility of the onset of delirium; a classifier creation means for creating a classifier that determines whether a patient has delirium or is likely to develop delirium based on the feature amounts extracted from the electroencephalograms of patients who have developed delirium or who may develop delirium, and the feature amounts extracted from the electroencephalograms of patients who have not developed delirium, based on the feature amounts; a determination means for determining whether the patient is experiencing delirium or is likely to experience delirium based on the feature extracted from the electroencephalogram of the patient and the classifier; A diagnostic support system comprising:

2. the feature extraction means extracts a plurality of frequency components from the electroencephalogram of a predetermined region of the brain, and extracts the frequency component of a predetermined frequency as the feature; 2. The diagnosis support system according to claim 1.

3. the classifier creation means selects a portion of the feature quantities in the process of creating the classifier, and creates the classifier based on the selected portion of the feature quantities.

3. The diagnosis support system according to claim 1 or 2.

4. the classifier creation means creates the classifier by performing multivariate analysis using the feature amounts as explanatory variables and whether or not delirium has developed or is likely to develop as a response variable.

3. The diagnosis support system according to claim 1 or 2.

5. The feature amount may include both a first feature amount whose value increases with the onset of delirium or the possibility of the onset of delirium, and a second feature amount whose value decreases with the onset of delirium or the possibility of the onset of delirium.

3. The diagnosis support system according to claim 1 or 2.

6. 3. The diagnostic support system according to claim 1, wherein the electroencephalogram is measured by a pair of electrodes consisting of a first electrode and a second electrode.

7. an acquisition means for acquiring a feature quantity extracted from an electroencephalogram and correlated with the onset of delirium or the possibility of the onset of delirium; a classifier that is created based on the feature amounts extracted from the electroencephalograms of patients who have developed delirium or who may develop delirium, and the feature amounts extracted from the electroencephalograms of patients who have not developed delirium, and that determines whether a patient has developed delirium or is likely to develop delirium based on the feature amounts; a determination means for determining whether the patient has developed delirium or is likely to develop delirium based on the feature amount extracted from the electroencephalogram of the patient to be diagnosed acquired by the acquisition means and the classifier; A diagnostic support system comprising:

8. 8. The diagnostic support system according to claim 7, wherein the feature amount is a plurality of frequency components extracted from electroencephalograms of a predetermined region of the brain.

9. 9. The diagnostic support system according to claim 7, wherein the electroencephalogram is measured by a pair of electrodes consisting of a first electrode and a second electrode.

10. a feature extraction step of extracting, from the electroencephalogram, a feature correlated with the onset of delirium or the possibility of the onset of delirium; a classifier creation step of creating a classifier that determines whether a patient has delirium or is likely to develop delirium based on the feature amounts extracted from the electroencephalograms of patients who have developed delirium or who may develop delirium, and the feature amounts extracted from the electroencephalograms of patients who have not developed delirium, based on the feature amounts; a determination step of determining whether the patient to be diagnosed has developed delirium or is likely to develop delirium based on the feature amount extracted from the electroencephalogram of the patient and the classifier; A diagnostic support method comprising:

11. a feature extraction function that extracts features that correlate with the onset of delirium or the possibility of the onset of delirium from the EEG; a classifier creation function that creates a classifier that determines whether a patient has delirium or is likely to develop delirium based on the feature amounts extracted from the electroencephalograms of patients who have developed delirium or who may develop delirium, and the feature amounts extracted from the electroencephalograms of patients who have not developed delirium, based on the feature amounts; a determination function that determines whether the patient is experiencing delirium or is likely to experience delirium based on the feature extracted from the electroencephalogram of the patient to be diagnosed and the classifier; A diagnostic support program characterized by causing a computer to realize the above.

12. a feature acquisition step of acquiring, from the electroencephalogram, a feature correlated with the onset of delirium or the possibility of the onset of delirium; The feature amount extracted from the electroencephalogram of a patient to be diagnosed; Based on the feature values ​​extracted from the EEG of a patient who has developed delirium or who may develop delirium, and the feature values ​​extracted from the EEG of a patient who has not developed delirium, a classifier is used to determine whether the patient has developed delirium or who may develop delirium from the feature values. determining whether the patient is experiencing or is at risk of experiencing delirium; A diagnostic support method comprising:

13. a feature acquisition function that acquires, from the electroencephalogram, feature values ​​that are correlated with the onset of delirium or the possibility of the onset of delirium; The feature amount extracted from the electroencephalogram of a patient to be diagnosed; Based on the feature values ​​extracted from the EEG of a patient who has developed delirium or who may develop delirium, and the feature values ​​extracted from the EEG of a patient who has not developed delirium, a classifier is used to determine whether the patient has developed delirium or who may develop delirium from the feature values. A determination function for determining whether the patient has developed delirium or is likely to develop delirium; A diagnostic support program characterized by causing a computer to realize the above.