Symptom prediction device and symptom prediction method

Abstract

To perform prediction on a symptom of a cardiac failure or a kidney failure.SOLUTION: A symptom prediction device 1 includes an acquisition unit 11 for acquiring voice information on a user's voice, and a prediction unit 12 for performing prediction on a symptom of the user's cardiac failure or kidney failure on the basis of the voice information acquired by the acquisition unit 11. The prediction unit 12 may perform prediction on the basis of voice change information on a change in the voice acquired on the basis of the voice information. The prediction unit 12 may perform prediction using a prediction model that performs prediction according to a degree of the voice change. The prediction unit 12 may perform prediction further on the basis of past voice change information. The prediction unit 12 may perform prediction on the basis of an acoustic feature amount acquired on the basis of the voice information.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] One aspect of the present disclosure relates to a symptom prediction device and a symptom prediction method for making predictions regarding symptoms of heart failure or renal failure of a user. [Background technology]

[0002] The following Patent Document 1 discloses an apparatus configured to receive voice of a subject suffering from a pulmonary disease associated with the accumulation of excess fluid, identify one or more voice-related parameters of the voice by analyzing the voice, assess the state of the pulmonary disease in response to the voice-related parameters, and generate an output indicative of the state of the pulmonary disease in response to the assessment. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Special Publication No. 2018-534026 Summary of the Invention [Problem to be solved by the invention]

[0004] Although the above device generates an output indicative of the state of lung disease, it is not able to make predictions regarding the onset of, for example, heart failure or renal failure, which would be desirable. [Means for solving the problem]

[0005] A symptom prediction device according to one aspect of the present disclosure includes an acquisition unit that acquires audio information related to a user's voice, and a prediction unit that makes a prediction regarding the user's symptoms of heart failure or renal failure based on the audio information acquired by the acquisition unit.

[0006] In such an aspect, predictions can be made regarding the condition of heart failure or renal failure. Effect of the Invention

[0007] According to one aspect of the present disclosure, predictions can be made regarding symptoms of heart failure or renal failure. [Brief description of the drawings]

[0008] [Figure 1] FIG. 2 is a diagram illustrating an example of a functional configuration of a symptom prediction device according to an embodiment. [Diagram 2] 11 is a flowchart illustrating an example of a process executed by the symptom prediction device according to the embodiment. [Diagram 3] FIG. 2 is a diagram illustrating an example of a functional configuration of a prediction unit of the symptom prediction device according to the first embodiment. [Figure 4] 5 is a flowchart showing an example of processing executed by a prediction unit of the symptom prediction device according to the first embodiment. [Diagram 5] FIG. 11 is a diagram illustrating an example of a functional configuration of a prediction unit of a symptom prediction device according to a second embodiment. [Figure 6] 13 is a flowchart showing an example of processing executed by a prediction unit of the symptom prediction device according to the second embodiment. [Figure 7] FIG. 13 is a diagram illustrating an example of a functional configuration of a prediction unit of a symptom prediction device according to a third embodiment. [Figure 8] 13 is a flowchart showing an example of processing executed by a prediction unit of the symptom prediction device according to the third embodiment. [Figure 9] 9 is a flowchart showing an example of detailed processing of step S34 in FIG. 8. [Figure 10] 10 is a flowchart showing another example of the detailed process of step S34 in FIG. 8. [Figure 11] FIG. 13 is a diagram illustrating an example of a functional configuration of a prediction unit of a symptom prediction device according to a fourth embodiment. [Figure 12] 13 is a flowchart showing an example of processing executed by a prediction unit of the symptom prediction device according to the fourth embodiment. [Figure 13] FIG. 2 is a diagram illustrating an example of a hardware configuration of a computer used in the symptom prediction device according to the embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0009] Hereinafter, the embodiments of the present disclosure will be described in detail with reference to the drawings. In the description of the drawings, the same elements are given the same reference numerals, and duplicated descriptions will be omitted. In addition, the embodiments of the present disclosure in the following description are specific examples of the present invention, and the present invention is not limited to these embodiments unless otherwise specified to limit the present invention.

[0010] Fig. 1 is a diagram showing an example of a functional configuration of a symptom prediction device 1 according to an embodiment. As shown in Fig. 1, the symptom prediction device 1 includes a storage unit 10 (storage unit), an acquisition unit 11 (acquisition unit), and a prediction unit 12 (prediction unit).

[0011] Each functional block of the symptom prediction device 1 (including sub-functional blocks (described later) contained in each functional block; the same applies below) is assumed to function within the symptom prediction device 1, but is not limited to this. For example, some of the functional blocks of the symptom prediction device 1 may function within a computer device different from the symptom prediction device 1 and connected to the symptom prediction device 1 through a network, while appropriately transmitting and receiving information with the symptom prediction device 1. Also, some of the functional blocks of the symptom prediction device 1 may be omitted, multiple functional blocks may be integrated into one functional block, or one functional block may be decomposed into multiple functional blocks.

[0012] Hereinafter, each function of the symptom prediction device 1 shown in FIG. 1 will be described.

[0013] The storage unit 10 stores any information. The storage unit 10 may store any information used in calculations in the symptom prediction device 1, or may store information related to the results of calculations in the symptom prediction device 1. The information stored by the storage unit 10 may be stored in advance, may be stored at any timing, or may be stored periodically. The information stored by the storage unit 10 may be referred to by each functional block of the symptom prediction device 1 as appropriate.

[0014] The acquisition unit 11 acquires any information. The acquisition unit 11 may acquire (receive) information from another device via a communication device 1004 (described later), may acquire information via an input device 1005 (described later), may acquire information stored by the storage unit 10, or may acquire information from another functional block of the symptom prediction device 1.

[0015] The acquiring unit 11 may cause the acquired information to be stored in the storage unit 10, or may output the information to other functional blocks of the symptom prediction device 1.

[0016] The acquisition unit 11 may acquire voice information related to a person's voice. In the embodiment, a "person" is described as a "user", but this is not limited thereto, and the "user" may be replaced with a "person" as appropriate. That is, the acquisition unit 11 may acquire voice information related to a user's voice. A user is a person who uses the symptom prediction device 1. A user ID (user identification information) that identifies the user is assigned to each user, and the symptom prediction device 1 may perform processing and storage of information for each user using the user ID. Processing and storage of information for each user using the user ID is performed by existing technology, and details are omitted as appropriate in the embodiment.

[0017] The voice information on the user's voice is any information on the user's voice. The voice information may include voice waveform data, which is waveform data indicating the user's voice. For example, the voice information may include voice waveform data obtained through a microphone when the user speaks a predetermined fixed phrase or when the user speaks freely. The voice information may include information on the user's voice and may further include a user ID for identifying the user. In this case, when processing the voice information, the symptom prediction device 1 may use the user ID included in the voice information to perform processing and storage of information for each user.

[0018] The acquisition unit 11 may cause the storage unit 10 to store the acquired voice information, or may output the information to the prediction unit 12 (including sub-functional blocks of the prediction unit 12).

[0019] The prediction unit 12 predicts the symptoms of the user's illness. In the embodiment, the "illness" is described as "heart failure or renal failure", but this is not limited thereto, and "heart failure or renal failure" may be replaced with "illness" as appropriate. That is, the prediction unit 12 predicts the symptoms of the user's heart failure or renal failure. The prediction unit 12 may predict the symptoms of the user's heart failure or renal failure based on any information. The prediction unit 12 may predict the symptoms of the user's heart failure or renal failure based on the user's voice information. The prediction unit 12 may predict the symptoms of the user's heart failure or renal failure based on the user's voice information acquired by the acquisition unit 11. The prediction unit 12 may predict the symptoms of the user's heart failure or renal failure based on the user's voice information input from the acquisition unit 11 or the user's voice information stored by the storage unit 10.

[0020] In the embodiments, the term "symptoms of cardiac failure or renal failure" is simply referred to as "symptoms" as appropriate. In the embodiments, the term "symptoms" may be appropriately read as "symptoms of cardiac failure or renal failure" or "symptoms of a disease."

[0021] The prediction performed by the prediction unit 12 may be a prediction of the presence or absence of a symptom. For example, the prediction unit 12 may predict the presence or absence of a symptom with a binary value of "0" or "1". In this case, "0" may indicate the absence of the symptom (almost absent, likely absent) and "1" may indicate the presence of the symptom (almost present, likely present). Or, conversely, "0" may indicate the presence of the symptom (almost present, likely present) and "1" may indicate the absence of the symptom (almost absent, likely absent). The above-mentioned binary value may be called a symptom prediction result (prediction result).

[0022] The prediction performed by the prediction unit 12 may be a prediction of the quantitative goodness or badness of the symptom. For example, the prediction unit 12 may indicate the quantitative goodness or badness of the symptom with a real number between "0" and "1". In this case, "0" indicates that the symptom is the best (complete absence of the symptom), and the closer to "0" the better the symptom (absence of the symptom), "1" indicates that the symptom is the worst (complete presence of the symptom), and the closer to "1" the worse the symptom (presence of the symptom). Or, conversely, "0" indicates that the symptom is the worst (complete presence of the symptom), and the closer to "0" the worse the symptom (presence of the symptom), and "1" indicates that the symptom is the best (complete absence of the symptom), and the closer to "1" the better the symptom (absence of the symptom). The above-mentioned real number between "0" and "1" may be called a symptom prediction result.

[0023] The prediction performed by the prediction unit 12 may be prediction based on words. For example, the prediction unit 12 may generate (and output) words such as "no symptoms", "symptoms are present", "symptoms have not changed", "symptoms have changed", "symptoms are getting better", "symptoms are getting worse", etc. The above-mentioned words may be called symptom prediction results.

[0024] When the prediction unit 12 is unable to perform the prediction appropriately, the prediction unit 12 may indicate this as a symptom prediction result. The prediction unit 12 may indicate that the prediction cannot be performed appropriately by a predetermined value, or may indicate this by the phrase "Prediction by voice cannot be performed appropriately", or the like.

[0025] The prediction unit 12 may output the prediction result (regarding symptoms) or may store it in the storage unit 10. For example, the prediction unit 12 may output (display) the prediction result to a user or the like via an output device 1006 (described later), or may output (transmit) the prediction result to another device or the like via a communication device 1004. The output result by the prediction unit 12 may be used as one of the decision-making materials for a doctor or the like to make a diagnosis, for example.

[0026] In the embodiments, "prediction of symptoms" is simply referred to as "prediction." In the embodiments, "prediction" may be read as "prediction of symptoms."

[0027] The prediction unit 12 may make a prediction based on voice change information related to a voice change obtained based on the voice information. The prediction unit 12 may calculate voice change information based on the voice information and make a prediction based on the calculated voice change information. The prediction unit 12 may obtain voice change information calculated by another functional block or device based on the voice information and make a prediction based on the obtained voice change information. The user's voice change information may be obtained based on the user's current (target timing) voice information and the user's past voice information (stored by the storage unit 10). The prediction unit 12 may make a prediction further based on past voice change information. That is, the prediction unit 12 may make a prediction based on the user's current (target timing) voice change information and the user's past voice change information (stored by the storage unit 10).

[0028] The voice change is a change in voice, such as age-related voice change, age-related voice disorder (hoarseness, rising pitch), etc. In the embodiment, "hoarseness" is mainly taken up as the "voice change", but is not limited to this. In the embodiment, "hoarseness" may be read as "voice change", "age-related voice change", "age-related voice disorder", "rising pitch", etc.

[0029] The voice change information is any information related to a voice change. For example, the voice change information may include information indicating the presence or absence of a voice change, or may include information indicating the degree of a voice change.

[0030] Here, we will explain the background. Patients with chronic heart failure at home measure data such as blood pressure, weight, and heart rate on a daily basis and report it to their doctors. Based on the reported data, doctors predict the patient's symptoms and decide whether or not to visit the hospital, but sometimes they overlook something. Patients with chronic renal failure also measure their own weight and other data every day and report it to their doctors, and the doctors adjust the timing of dialysis based on weight gain or loss, but sometimes it is difficult to make a judgment.

[0031] Chronic heart failure, chronic renal failure, and other diseases often cause changes in the amount of water in the body. The amount of water in the body affects speech. The effect on speech is reflected in voice changes. Based on the above background, the prediction unit 12 (or the symptom prediction device 1) focuses on the (direct) link between heart failure and renal failure and voice changes, and makes predictions based on voice change information. This allows the prediction unit 12 (or the symptom prediction device 1) to achieve high prediction performance for users experiencing voice changes. Note that the device disclosed in the above-mentioned Patent Document 1 cannot determine the presence or absence of symptoms when the subject's voice includes voice changes such as hoarseness.

[0032] The progression stages of heart failure and renal failure will be explained.

[0033] Heart failure can be divided into stages A, B, C, and D. In stage A, symptoms include hypertension, diabetes, and arteriosclerosis. In stage B, symptoms include ischemic heart disease, left ventricular remodeling (left ventricular hypertrophy and reduced ejection fraction), and asymptomatic valvular disease. From stage B onwards, sudden death may occur. In stage C, symptoms include acute heart failure, acute aggravation of chronic heart failure, and recurrence of (acute heart failure). From stage C onwards, it becomes chronic heart failure. In stage D, symptoms include acute aggravation of chronic heart failure and recurrence of (acute heart failure). In stage D, heart failure becomes difficult to treat. As the patient progresses from stage A to stage D, physical function declines over time. Stages A and B of heart failure are simply high-risk groups, and it is only from stage C onwards that symptoms of heart failure actually begin to appear. It can also be considered that the patient enters stage C as developing heart failure in the latter half of stage B. For this reason, the prediction unit 12 (or the symptom prediction device 1) may target stage C of heart failure.

[0034] Renal failure can be divided into stage 1, stage 2, stage 3, stage 4, and stage 5. Each stage is classified according to kidney function. In stages 1 and 2, there are no subjective symptoms. In stages 3, 4, and 5, there are subjective symptoms, and treatment methods include lifestyle improvement, dietary therapy, and drug therapy. In stage 5, dialysis is added as a treatment method. Specifically, in stage 5, renal function is further reduced and the body is no longer able to effectively excrete water accumulated in the body, making dialysis treatment necessary. For this reason, the prediction unit 12 (or the symptom prediction device 1) may target stage 5 of renal failure.

[0035] Heart failure and renal failure have in common the phenomenon that "fluid accumulates in the body, causing swelling." The prediction unit 12 (or the symptom prediction device 1) captures "fluid accumulates in the body, causing swelling" from voice change information.

[0036] The prediction unit 12 may perform prediction based on the degree of voice change. The prediction unit 12 may perform prediction using a prediction model that performs prediction according to the degree of voice change. For example, a plurality of prediction models that perform prediction according to the degree of voice change for each degree of voice change may be prepared in advance (prestored in advance by the storage unit 10), and the prediction unit 12 may select a prediction model according to the degree of voice change based on the current (target timing) voice information from the plurality of prediction models, and perform prediction using the selected prediction model.

[0037] The prediction unit 12 may make a prediction based on an acoustic feature obtained based on the voice information. The prediction unit 12 may make a prediction based on an acoustic feature obtained based on the voice information for each prediction of the prediction unit 12 described above. As an example, the prediction unit 12 may make a prediction based on voice change information obtained based on the voice information and the acoustic feature obtained based on the voice information.

[0038] Next, an example of processing executed by the symptom prediction device 1 will be described with reference to Fig. 2. Fig. 2 is a flowchart showing an example of processing (symptom prediction method) executed by the symptom prediction device 1.

[0039] First, the acquisition unit 11 acquires voice information related to the user's voice (step S1, acquisition step). Next, the prediction unit 12 predicts the symptoms of the user's heart failure or renal failure based on the voice information acquired in step S1 (step S2, prediction step).

[0040] Next, the prediction unit 12 of the symptom prediction device 1 according to some embodiments will be described in detail with reference to Figs. 3 to 12. Note that a device or functional block having a reference symbol with an alphabetical letter added at the end (e.g., the prediction unit 12A) may have the functions of a device or functional block having a reference symbol with no alphabetical letter added (e.g., the prediction unit 12) and a device or functional block having a reference symbol with another alphabetical letter added (e.g., the prediction unit 12B).

[0041] [First embodiment] 3 is a diagram showing an example of the functional configuration of a prediction unit 12A of a symptom prediction device 1A according to the first embodiment. The symptom prediction device 1A is similar to the symptom prediction device 1, and includes a storage unit 10A (not shown), an acquisition unit 11A (not shown), and a prediction unit 12A as functional blocks having similar functions to the storage unit 10, acquisition unit 11, and prediction unit 12 described thus far. The difference between the prediction unit 12A and the prediction unit 12 will be described below.

[0042] As shown in FIG. 3, the prediction unit 12A includes, as sub-functional blocks that are lower-level functional blocks, a symptom prediction use acoustic feature extraction unit 120A, a hoarseness determination use acoustic feature extraction unit 121A, and a symptom prediction unit 125A.

[0043] The symptom prediction acoustic feature extraction unit 120A extracts acoustic features (acoustic features) indicated by a user voice (another name for voice information; the same applies below) based on the user voice. The extracted features are called symptom prediction acoustic features because they are used to predict symptoms. Specifically, the symptom prediction acoustic feature extraction unit 120A inputs the user voice (from the storage unit 10A or the acquisition unit 11A, etc.), extracts symptom prediction acoustic features from the input user voice, and may output the extracted symptom prediction acoustic features (to the symptom prediction unit 125A, etc.) or may store them in the storage unit 10A.

[0044] The symptom prediction acoustic feature extraction unit 120A extracts the symptom prediction acoustic feature by combining existing techniques. For example, existing extraction tools such as sptk and parselmouth may be used. The features extracted by the symptom prediction acoustic feature extraction unit 120A may be fundamental frequency, HNR (Harmonic-to-Noise Ratio), formant, MFCC (Mel-Frequency Cepstrum Coefficient), Jitter, Shimmer, etc.

[0045] The symptom prediction acoustic feature may be a one-dimensional value. The symptom prediction acoustic feature is a value obtained by averaging one piece of speech waveform data (user speech) in the time series direction, and may be held by the number of features. The symptom prediction acoustic feature may be a set of pairs of the type of feature and the value of the feature of the type, for example, [feature 1: 0.3, feature 2: -0.2, ..., feature x: 1.4]. In this case, it indicates that the value of feature 1 is "0.3", the value of feature 2 is "-0.2", ..., and the value of feature x is "1.4". Note that the feature and the difference of the feature in the asymptomatic state that will be described later in the third embodiment may also be in the same format.

[0046] The symptom prediction acoustic feature may be an n-dimensional (n: integer of 2 or more) value, such as a time series of feature values. When the feature values ​​are used for judgment, they are converted to one dimension. For example, the feature values ​​may be [feature value 1: [0.3, 0.1, 0.4, 0.2], feature value 2: [-0.2, 0.0, -0.3, 0.1], ..., feature value x: [1.4, 1.3, 1.6, 1.5]]. In this case, the values ​​of feature value 1 are "0.3", "0.1", "0.4", "0.2" in chronological order, the values ​​of feature value 2 are "-0.2", "0.0", "-0.3", "0.1" in chronological order, ..., and the values ​​of feature value x are "1.4", "1.3", "1.6", "1.5". After converting to 1 dimension, it becomes [0.3, 0.1, 0.4, 0.2, -0.2, 0.0, -0.3, 0.1, …, 1.4, 1.3, 1.6, 1.5].

[0047] In the process of extracting the acoustic feature for symptom prediction, the acoustic feature for symptom prediction extraction unit 120A may extract a difference value compared with the acoustic feature for base as the acoustic feature for final symptom prediction. For example, the voice information or the acoustic feature for symptom prediction when the user is healthy may be stored in advance by the storage unit 10 and used in the process of extraction. In addition, the acoustic feature for symptom prediction extraction unit 120A may use an average base value extracted from the voice information or the acoustic feature for symptom prediction when a plurality of other users are healthy, instead of the data when the user is healthy, to obtain the difference from the base value. The extracted acoustic feature for symptom prediction, the base (for example, when healthy) feature, and the difference feature are all one-dimensional values, and may be provided for each feature. Since there is a possibility that one of the acoustic features has a different range of values ​​for men and women, it may be difficult to determine whether it is just getting larger or is getting larger due to the influence of gender. In such a case, the accuracy of the final prediction is improved by using the difference.

[0048] The acoustic feature quantity for hoarseness determination extraction unit 121A extracts acoustic features indicated by the user voice based on the user voice. The extracted features are called acoustic feature quantities for hoarseness determination because they are used to determine hoarseness. Specifically, the acoustic feature quantity for hoarseness determination extraction unit 121A may input the user voice (from the storage unit 10A or the acquisition unit 11A, etc.), extract acoustic feature quantities for hoarseness determination from the input user voice, and output the extracted acoustic feature quantities for hoarseness determination (to the symptom prediction unit 125A, etc.) or may store them in the storage unit 10A.

[0049] The method of extracting the acoustic feature for hoarseness determination by the acoustic feature for hoarseness determination extraction unit 121A may be a calculation based on the definition of CPP (Cepstrum Peak Prominence), a calculation based on the definition of CSID (Cepstral Spectral Index of Dysphonia), or a calculation based on any other definition (e.g., related to voice change). In addition, the fundamental frequency extracted by the above-mentioned existing extraction tool may be used in the extraction process in the calculation. The acoustic feature for hoarseness determination extracted by the acoustic feature for hoarseness determination extraction unit 121A may be the value of the calculation result based on the definition of CPP itself or a value based on the said value, the value of the calculation result based on the definition of CSID itself or a value based on the said value, the value of the calculation result based on any other definition (e.g., related to voice change) itself or a value based on the said value, or at least two or more of them.

[0050] The symptom prediction unit 125A predicts symptoms based on the acoustic feature for symptom prediction and the acoustic feature for hoarseness determination. Specifically, the symptom prediction unit 125A inputs the acoustic feature for symptom prediction and the acoustic feature for hoarseness determination (from the storage unit 10A or the acoustic feature for symptom prediction extraction unit 120A and the acoustic feature for hoarseness determination extraction unit 121A, respectively), applies the input acoustic feature for symptom prediction and the acoustic feature for hoarseness determination to a symptom prediction model, and may output a symptom prediction result obtained as a result of the application, or may cause the storage unit 10A to store the result.

[0051] The symptom prediction model is a prediction model trained based on training data consisting of sets of acoustic features for symptom prediction and acoustic features for hoarseness determination, and symptom prediction results. When the acoustic features for symptom prediction and the acoustic features for hoarseness determination are applied (input), the symptom prediction model outputs a symptom prediction result.

[0052] FIG. 4 is a flowchart showing an example of a process executed by the prediction unit 12A of the symptom prediction device 1A according to the first embodiment. First, the prediction unit 12A inputs a user voice (step S10). Next, the symptom prediction acoustic feature extraction unit 120A extracts a symptom prediction acoustic feature based on the user voice input in step S10 (step S11). Next, the hoarseness determination acoustic feature extraction unit 121A extracts a hoarseness determination acoustic feature based on the user voice input in step S10 (step S12). Next, the symptom prediction unit 125A applies the symptom prediction acoustic feature extracted in step S11 and the hoarseness determination acoustic feature extracted in step S12 to a symptom prediction model (using the symptom prediction model), and outputs (predicts) a symptom prediction result obtained as a result of the application (step S13).

[0053] In the flowchart shown in FIG. 4, the order of steps S11 and S12 may be reversed.

[0054] [Second embodiment] 5 is a diagram showing an example of a functional configuration of a prediction unit 12B of a symptom prediction device 1B according to the second embodiment. The symptom prediction device 1B is similar to the symptom prediction device 1, and includes a storage unit 10B (not shown), an acquisition unit 11B (not shown), and a prediction unit 12B as functional blocks having similar functions to the storage unit 10, acquisition unit 11, and prediction unit 12 described thus far. The prediction unit 12B may have similar functions to the prediction unit 12A of the first embodiment. The following describes the difference between the prediction unit 12B and the prediction unit 12A.

[0055] As shown in FIG. 5, the prediction unit 12B includes, as sub-functional blocks, an acoustic feature for symptom prediction extraction unit 120B, an acoustic feature for hoarseness determination extraction unit 121B, a hoarseness degree determination unit 122B, a symptom prediction model selection unit 123B, and a symptom prediction unit 125B.

[0056] The symptom prediction acoustic feature extraction unit 120B may output the extracted symptom prediction acoustic feature to the symptom prediction unit 125B or the like, or may store it in the storage unit 10B.

[0057] The hoarseness determining acoustic feature extracting unit 121B may output the extracted hoarseness determining acoustic feature to the hoarseness level determining unit 122B and the symptom predicting unit 125B, or may store it in the storage unit 10B.

[0058] The hoarseness level determination unit 122B determines the degree of voice change. The hoarseness level determination unit 122B may determine a hoarseness level, which is the degree of hoarseness of the user. The hoarseness level determination unit 122B may determine the hoarseness level of the user based on the acoustic feature for hoarseness determination. The hoarseness level determination unit 122B may determine the hoarseness level of the user based on the acoustic feature for hoarseness determination stored by the storage unit 10B or input from the acoustic feature for hoarseness determination extraction unit 121B.

[0059] The hoarseness level determination unit 122B may determine whether the acoustic feature for hoarseness determination is less than or equal to a predetermined threshold. For example, when the acoustic feature for hoarseness determination that is input is a value between "0" and "1", the hoarseness level determination unit 122B may determine whether the acoustic feature for hoarseness determination is lower (lower) than the threshold value "0.5" (same unit as the numerical value of the acoustic feature for hoarseness determination) (low hoarseness level) or higher (high hoarseness level) than the threshold value. The hoarseness level determination unit 122B may perform three or more case classifications instead of two cases, that is, less than or equal to the threshold value. The hoarseness level determination unit 122B may indicate the hoarseness level as a real number between "0" and "1" (which may be in the same unit as the numerical value of the acoustic feature for hoarseness determination, or may be in a different unit). In this case, "0" may indicate the lowest degree of hoarseness, with the closer to "0" the lower the degree of hoarseness, "1" may indicate the highest degree of hoarseness, and the closer to "1" the higher the degree of hoarseness. Or vice versa, "0" may indicate the highest degree of hoarseness, with the closer to "0" the higher the degree of hoarseness, and "1" may indicate the lowest degree of hoarseness, and the closer to "1" the lower the degree of hoarseness.

[0060] The hoarseness level determination unit 122B may determine the hoarseness level from two values, CPP and CSID. For example, the hoarseness level determination unit 122B may determine the average of the hoarseness levels calculated from the respective indices of CPP and CSID as the final hoarseness level. When determining the hoarseness level, the hoarseness level determination unit 122B may take into consideration the values ​​of indices other than CPP and CSID (e.g., acoustic features that may be related to hoarseness). In that case, the hoarseness level determination unit 122B may determine the hoarseness level by taking a weighted average in which the values ​​of CPP and CSID are weighted heavily and other values ​​(e.g., acoustic features, etc.) are weighted lightly.

[0061] The hoarseness degree determining section 122B may output the result of the hoarseness degree determination as a hoarseness degree determination result to the symptom prediction model selecting section 123B and the symptom predicting section 125B, or may store it in the storage section 10B.

[0062] When hoarseness is present, the occurrence of hoarseness itself may indicate a worsening of symptoms, and therefore the hoarseness level determination unit 122B may determine whether symptoms are worsening based on the determined (calculated) hoarseness level. In this case, the result of the determination of the worsening of symptoms may be output (by the prediction unit 12B) as a symptom prediction result (rather than as a hoarseness level determination result). In addition, when the hoarseness level is too high (higher than a predetermined threshold), it is expected that voice changes will be difficult to see, so the hoarseness level determination unit 122B may output, for example, a statement such as "Prediction by voice cannot be performed appropriately" as a symptom prediction result. In addition, when the hoarseness level determination unit 122B determines whether symptoms are worsening and outputs it (by the prediction unit 12B) as a symptom prediction result, the processing of the symptom prediction acoustic feature extraction unit 120B, the processing of the symptom prediction model selection unit 123B, or the processing of the symptom prediction unit 125B may be omitted. In addition, when the hoarseness level is too high (higher than a predetermined threshold value) and the hoarseness level determination unit 122B outputs it as a symptom prediction result, the processing of the symptom prediction acoustic feature extraction unit 120B may be omitted, the processing of the symptom prediction model selection unit 123B may be omitted, and the processing of the symptom prediction unit 125B may be omitted.

[0063] The symptom prediction model selection unit 123B selects a symptom prediction model to be used in the symptom prediction unit 125B. The symptom prediction model selection unit 123B may select a symptom prediction model to be used in the symptom prediction unit 125B based on the hoarseness level judgment result input from the hoarseness level judgment unit 122B. For example, in the case where a "symptom prediction model (high hoarseness level)" which is a symptom prediction model to be used when the hoarseness level is high (above a predetermined threshold) and a "symptom prediction model (low hoarseness level)" which is a symptom prediction model to be used when the hoarseness level is low (below a predetermined threshold) are stored in advance by the storage unit 10B, the symptom prediction model selection unit 123B may select the "symptom prediction model (high hoarseness level)" when the hoarseness level judgment result indicates a high hoarseness level, and may select the "symptom prediction model (low hoarseness level)" when the hoarseness level judgment result indicates a low hoarseness level.

[0064] The "symptom prediction model (high hoarseness)" is a prediction model trained based on training data consisting of pairs of symptom prediction acoustic features when the hoarseness level is high (above a predetermined threshold) and symptom prediction results. The "symptom prediction model (high hoarseness)" outputs symptom prediction results when the symptom prediction acoustic features are applied (input). The "symptom prediction model (high hoarseness)" is a prediction model that outputs more accurate symptom prediction results when the hoarseness level is high.

[0065] The "symptom prediction model (low hoarseness)" is a prediction model trained based on training data consisting of pairs of symptom prediction acoustic features when the hoarseness level is low (below a predetermined threshold) and symptom prediction results. The "symptom prediction model (low hoarseness)" outputs symptom prediction results when the symptom prediction acoustic features are applied (input). The "symptom prediction model (low hoarseness)" is a prediction model that outputs more accurate symptom prediction results when the hoarseness level is low.

[0066] 5, for ease of understanding, the "symptom prediction model (high hoarseness)" and the "symptom prediction model (low hoarseness)" are illustrated as being included in the symptom prediction unit 125B, but in reality, the "symptom prediction model (high hoarseness)" and the "symptom prediction model (low hoarseness)" are stored by the storage unit 10B. Note that the symptom prediction unit 125B may store symptom prediction models such as the "symptom prediction model (high hoarseness)" and the "symptom prediction model (low hoarseness)" internally (not in the storage unit 10B).

[0067] The symptom prediction model selection unit 123B may output a model ID (model identification information) for identifying the selected symptom prediction model to the symptom prediction unit 125B or the like, or may cause the storage unit 10B to store it.

[0068] The symptom prediction unit 125B predicts symptoms based on the symptom prediction acoustic feature and the hoarseness level judgment result (hoarseness level). The symptom prediction unit 125B may predict symptoms based on the symptom prediction acoustic feature and a model ID. Specifically, the symptom prediction unit 125B inputs the symptom prediction acoustic feature (from the storage unit 10B or the symptom prediction acoustic feature extraction unit 120B) and inputs the model ID (from the storage unit 10B or the symptom prediction model selection unit 123B), applies the input symptom prediction acoustic feature to a symptom prediction model indicated by the input model ID, and may output a symptom prediction result obtained as a result of the application, or may store the symptom prediction result in the storage unit 10B.

[0069] FIG. 6 is a flowchart showing an example of a process executed by the prediction unit 12B of the symptom prediction device 1B according to the second embodiment. First, the prediction unit 12B inputs a user voice (step S20). Next, the symptom prediction acoustic feature extraction unit 120B extracts a symptom prediction acoustic feature based on the user voice input in step S20 (step S21). Next, the hoarseness determination acoustic feature extraction unit 121B extracts a hoarseness determination acoustic feature based on the user voice input in step S20 (step S22). Next, the hoarseness level determination unit 122B determines the hoarseness level based on the hoarseness level determination acoustic feature extracted in step S22, and outputs the hoarseness level determination result (step S23). Next, the symptom prediction model selection unit 123B selects a symptom prediction model based on the hoarseness level determination result output in step S23, and outputs the model ID of the selected symptom prediction model (step S24). Next, the symptom prediction unit 125B applies the acoustic features for symptom prediction extracted in step S21 to the symptom prediction model indicated by the model ID output in step S24 (using the symptom prediction model), and outputs (predicts) the symptom prediction result obtained as a result of the application (step S25).

[0070] In the flowchart shown in FIG. 6, step S21 may be executed at any time after step S20 and before step S25.

[0071] [Third embodiment] 7 is a diagram showing an example of a functional configuration of a prediction unit 12C of a symptom prediction device 1C according to the third embodiment. The symptom prediction device 1C is similar to the symptom prediction device 1, and includes a storage unit 10C, an acquisition unit 11C (not shown), and a prediction unit 12C as functional blocks having similar functions to the storage unit 10, acquisition unit 11, and prediction unit 12 described thus far. The prediction unit 12C may have similar functions to the prediction unit 12A of the first embodiment or the prediction unit 12B of the second embodiment. The difference between the prediction unit 12C and the prediction unit 12B will be described below.

[0072] First, the symptom prediction unit 125C will be described. When the symptom prediction result outputted by the symptom prediction unit 125C indicates no symptoms, the symptom prediction unit 125C causes the storage unit 10C to store the hoarseness level (the hoarseness level obtained at the present time (target timing) by the prediction unit 12) which is the hoarseness level judgment result inputted from the hoarseness level judgment unit 122C, the acoustic feature for hoarseness judgment (the acoustic feature for hoarseness judgment obtained at the present time (target timing) by the prediction unit 12) inputted from the acoustic feature for hoarseness judgment extraction unit 121C, and the acoustic feature for symptom prediction (the acoustic feature for hoarseness judgment obtained at the present time (target timing) by the prediction unit 12) inputted from the acoustic feature for symptom prediction extraction unit 120C. That is, when the symptom prediction result indicates no symptoms every time of prediction, the prediction unit 12C causes the storage unit 10C to store the hoarseness level, the acoustic feature for hoarseness judgment, and the acoustic feature for symptom prediction.

[0073] When extracting the acoustic feature quantity for symptom prediction, the acoustic feature quantity for symptom prediction extraction unit 120C acquires the acoustic feature quantity for symptom prediction when the user is asymptomatic stored by the storage unit 10C, and extracts the acoustic feature quantity for symptom prediction based on the acquired acoustic feature quantity for symptom prediction when the user is asymptomatic. The acoustic feature quantity for symptom prediction acquired may be an average value of the acoustic feature quantity for symptom prediction when the user is asymptomatic stored by the storage unit 10C. The acoustic feature quantity for symptom prediction extraction unit 120C may extract, for example, a difference (variation) between the acoustic feature quantity for symptom prediction indicated by the user's voice and the acoustic feature quantity for symptom prediction when the user is asymptomatic, as the final acoustic feature quantity for symptom prediction. The reason for taking the difference is that some acoustic features are influenced by individual characteristics.

[0074] When extracting the acoustic feature quantity for hoarseness determination, the acoustic feature quantity for hoarseness determination extraction unit 121C acquires the hoarseness level and the acoustic feature quantity for hoarseness determination of the user when the user has no symptoms stored by the storage unit 10C, and extracts the acoustic feature quantity for hoarseness determination based on the acquired acoustic feature quantity for hoarseness determination when the user has no symptoms. The acquired acoustic feature quantity for hoarseness determination when the user has no symptoms may be an average value of the hoarseness level stored by the storage unit 10C. The acquired acoustic feature quantity for hoarseness determination when the user has no symptoms may be an average value of the acoustic feature quantity for hoarseness determination when the user has no symptoms stored by the storage unit 10C. The acoustic feature quantity for hoarseness determination extraction unit 121C may extract, for example, a difference (variation) between the acoustic feature quantity for hoarseness determination indicated by the user's voice and the acoustic feature quantity for hoarseness determination of the user when the user has no symptoms, as the final acoustic feature quantity for hoarseness determination. The reason for taking the difference is that some acoustic features are influenced by individual characteristics. The hoarseness determination acoustic feature extraction unit 121C further outputs the acquired hoarseness level in the asymptomatic state to the hoarseness level determination unit 122C.

[0075] The hoarseness level determination unit 122C further determines whether or not there is chronic hoarseness based on the determined hoarseness level and the hoarseness level at an asymptomatic time input from the hoarseness determination acoustic feature extraction unit 121C (or acquired from the storage unit 10C). Specifically, the hoarseness level determination unit 122C extracts and compares the difference between the determined hoarseness level and the hoarseness level at an asymptomatic time input from the hoarseness level determination unit 122C (or acquired from the storage unit 10C) to determine whether or not there is chronic hoarseness. The hoarseness level determination unit 122C further includes the determination result of whether or not there is chronic hoarseness in the hoarseness level determination result and outputs it to the symptom prediction model selection unit 123C.

[0076] The symptom prediction model selection unit 123C may determine that there is no hoarseness if the hoarseness level is less than a predetermined threshold based on the hoarseness level judgment result input from the hoarseness level judgment unit 122C, and may determine that there is hoarseness if the hoarseness level is equal to or greater than the predetermined threshold. The symptom prediction model selection unit 123C may transition as follows based on the hoarseness level judgment result input from the hoarseness level judgment unit 122C. (1) If you have chronic hoarseness and the hoarseness level is high, select the "Symptom Prediction Model (High Hoarseness)." (2) If there is chronic hoarseness and the degree of hoarseness is low, a statement such as "symptoms have changed" is output as a symptom prediction result. (3) If there is no chronic hoarseness and the degree of hoarseness is high, a statement such as "symptoms have changed" is output as a symptom prediction result. (4) If there is no chronic hoarseness and the degree of hoarseness is low, select the "Symptom Prediction Model (Low Hoarseness)."

[0077] The reason why the symptom prediction result is output as a statement such as "symptoms have changed" is because it is not possible to determine whether the symptoms are good or bad. Also, if the degree of hoarseness is too high (above a certain threshold), it is expected that changes in the voice will be difficult to see, so the symptom prediction result may be output as a statement such as "prediction based on voice cannot be performed appropriately."

[0078] 8 is a flowchart showing an example of a process executed by the prediction unit 12C of the symptom prediction device 1C according to the third embodiment. First, the prediction unit 12C inputs a user voice (step S30). Next, the symptom prediction acoustic feature extraction unit 120C extracts a symptom prediction acoustic feature based on the user voice input in step S30 and the symptom prediction acoustic feature stored by the storage unit 10C when the user is asymptomatic (step S31). Next, the hoarseness determination acoustic feature extraction unit 121C extracts a hoarseness determination acoustic feature based on the user voice input in step S30 and the hoarseness level and hoarseness determination acoustic feature stored by the storage unit 10C when the user is asymptomatic (step S32). Next, the hoarseness level determination unit 122C determines the hoarseness level based on the hoarseness determination acoustic feature extracted in step S32, and further determines the presence or absence of chronic hoarseness based on the asymptomatic hoarseness level stored by the storage unit 10C, and outputs the hoarseness level determination result (step S33). Next, the symptom prediction model selection unit 123C selects a symptom prediction model based on the hoarseness level determination result output in step S33, and outputs the model ID of the selected symptom prediction model (step S34).

[0079] Next, the symptom prediction unit 125C applies the symptom prediction acoustic feature extracted in step S31 to the symptom prediction model indicated by the model ID output in step S34 (using the symptom prediction model), and outputs (predicts) the symptom prediction result obtained as a result of the application (step S35). Next, if the symptom prediction result output by the symptom prediction unit 125C in step S35 indicates no symptoms, the hoarseness level, which is the hoarseness level determination result output in step S33, the hoarseness determination acoustic feature output in step S32, and the symptom prediction acoustic feature output in step S31 are stored in the storage unit 10C (step S36).

[0080] In the flowchart shown in FIG. 8, step S31 may be executed at any time after step S30 and before step S35.

[0081] Fig. 9 is a flow chart showing an example of detailed processing of step S34 in Fig. 8. First, the symptom prediction model selection unit 123C judges whether or not there is chronic hoarseness based on the hoarseness level judgment result outputted in step S33 (step S40). If it is judged that there is chronic hoarseness in step S40 (step S40: Yes), the symptom prediction model selection unit 123C judges whether the hoarseness level is high or low based on the hoarseness level judgment result (step S41). If it is judged that the hoarseness level is high in step S41 (step S41: High), the symptom prediction model selection unit 123C selects the "symptom prediction model (high hoarseness level)" (step S42).

[0082] On the other hand, if it is determined that the hoarseness level is not present in step S40 (step S40: no), the symptom prediction model selection unit 123C determines whether the hoarseness level is high or low based on the hoarseness level determination result (step S43). If it is determined that the hoarseness level is high in step S43 (step S43: high), or if it is determined that the hoarseness level is low in step S41 (step S41: low), the symptom prediction model selection unit 123 outputs a phrase such as "symptoms have changed" as a symptom prediction result (step S44). On the other hand, if it is determined that the hoarseness level is low in step S43 (step S43: low), the symptom prediction model selection unit 123C selects "symptom prediction model (low hoarseness level)" (step S45). Note that, if the process proceeds to step S44, the process of the subsequent step S35 may be omitted, and the process of the subsequent step S36 may be omitted.

[0083] Fig. 10 is a flow chart showing another example of the detailed process of step S34 in Fig. 8. First, the symptom prediction model selection unit 123C judges whether the degree of hoarseness is high or low based on the hoarseness degree judgment result outputted in step S33 (step S50). If it is judged to be high in step S50 (step S50: high), the symptom prediction model selection unit 123C judges whether there is chronic hoarseness or not based on the hoarseness degree judgment result (step S51). If it is judged to be present in step S51 (step S51: yes), the symptom prediction model selection unit 123C selects the "symptom prediction model (high hoarseness degree)" (step S52).

[0084] On the other hand, if it is determined that the hoarseness level is low in step S50 (step S50: low), the symptom prediction model selection unit 123C determines whether or not there is chronic hoarseness based on the hoarseness level determination result (step S53). If it is determined that there is chronic hoarseness in step S53 (step S53: yes), or if it is determined that there is no chronic hoarseness in step S51 (step S51: no), the symptom prediction model selection unit 123 outputs a phrase such as "symptoms have changed" as a symptom prediction result (step S54). On the other hand, if it is determined that there is no chronic hoarseness in step S53 (step S53: no), the symptom prediction model selection unit 123C selects "symptom prediction model (low hoarseness level)" (step S55). Note that, when proceeding to the process of step S54, the process of the subsequent step S35 may be omitted, and the process of the subsequent step S36 may be omitted.

[0085] [Fourth embodiment] 11 is a diagram showing an example of a functional configuration of a prediction unit 12D of a symptom prediction device 1D according to the fourth embodiment. The symptom prediction device 1D is similar to the symptom prediction device 1, and includes a storage unit 10D (not shown), an acquisition unit 11D (not shown), and a prediction unit 12D as functional blocks having similar functions to the storage unit 10, acquisition unit 11, and prediction unit 12 described above. The prediction unit 12D may have similar functions to the prediction unit 12A of the first embodiment, the prediction unit 12B of the second embodiment, or the prediction unit 12C of the third embodiment. The difference between the prediction unit 12D and the prediction unit 12C will be described below.

[0086] As shown in FIG. 11, the prediction unit 12D includes, as sub-functional blocks, a symptom prediction acoustic feature extraction unit 120D, a hoarseness determination acoustic feature extraction unit 121D, a hoarseness level determination unit 122D, a feature selection unit 124D, and a symptom prediction unit 125D.

[0087] The symptom prediction acoustic feature extraction unit 120D may output the extracted symptom prediction acoustic feature to the feature selection unit 124D or the like, or may store it in the storage unit 10D.

[0088] The hoarseness level determination unit 122D may output the hoarseness level, which is the result of the hoarseness level determination, to the feature amount selection unit 124D or the like, or may store it in the storage unit 10D.

[0089] The feature selection unit 124D selects a predetermined acoustic feature group based on the degree of hoarseness, and outputs to the symptom prediction unit 125D acoustic features for model construction, including acoustic features included in the selected acoustic feature group among the symptom prediction acoustic features, and a model ID of a symptom prediction model corresponding to the selected acoustic feature group. The storage unit 10D stores a symptom prediction model for each acoustic feature group in advance. For example, the storage unit 10D may store a "symptom prediction model (acoustic feature group 1)" which is a symptom prediction model corresponding to acoustic feature group 1, a "symptom prediction model (acoustic feature group 2)" which is a symptom prediction model corresponding to acoustic feature group 2, and so on. For example, the degree of hoarseness may be associated with a group of acoustic features in advance (for example, if the degree of hoarseness is between "0" and "0.5", acoustic feature set 1 may correspond, if the degree of hoarseness is between "0.5" and "1", acoustic feature set 2 may correspond, etc.), and the feature selection unit 124D may select a group of acoustic features based on the degree of hoarseness and the association.

[0090] A "symptom prediction model (acoustic feature group n)" (n is an integer equal to or greater than 1) is a prediction model trained based on training data consisting of pairs of acoustic features included in acoustic feature group n and symptom prediction results. A "symptom prediction model (acoustic feature group n)" outputs a symptom prediction result when acoustic features are applied (input). A "symptom prediction model (acoustic feature group n)" is a prediction model that can output more accurate symptom prediction results when acoustic features included in acoustic feature group n are applied.

[0091] The feature selection unit 124D, for example, can exclude acoustic features that are easily affected by hoarseness and select a symptom prediction model trained only with other acoustic features to predict symptoms, thereby preventing adverse effects on symptom prediction and improving prediction performance.

[0092] The symptom prediction unit 125D may apply the acoustic features included in the acoustic features for model construction input from the feature selection unit 124D to a symptom prediction model indicated by the model ID included in the acoustic features for model construction, and output the symptom prediction result obtained as a result of the application, or may store it in the storage unit 10D.

[0093] 11, for ease of understanding, the "symptom prediction model (acoustic feature group 1)" and the "symptom prediction model (acoustic feature group 2)" are illustrated as being included in the symptom prediction unit 125D, but in reality, the "symptom prediction model (acoustic feature group 1)" and the "symptom prediction model (acoustic feature group 2)" are stored by the storage unit 10D. Note that the symptom prediction unit 125D may store symptom prediction models such as the "symptom prediction model (acoustic feature group 1)" and the "symptom prediction model (acoustic feature group 2)" internally (not in the storage unit 10D).

[0094] 12 is a flowchart showing an example of a process executed by the prediction unit 12D of the symptom prediction device 1D according to the fourth embodiment. First, the prediction unit 12D inputs a user voice (step S60). Next, the symptom prediction acoustic feature extraction unit 120D extracts a symptom prediction acoustic feature based on the user voice input in step S60 (step S61). Next, the hoarseness determination acoustic feature extraction unit 121D extracts a hoarseness determination acoustic feature based on the user voice input in step S60 (step S62). Next, the hoarseness level determination unit 122D determines the hoarseness level based on the hoarseness determination acoustic feature extracted in step S62, and outputs the hoarseness level determination result (step S63). Next, the feature selection unit 124D selects a predetermined acoustic feature group based on the hoarseness level judgment result output in S63, and outputs a model construction acoustic feature including an acoustic feature included in the selected acoustic feature group from the symptom prediction acoustic features extracted in S61 and a model ID of a symptom prediction model corresponding to the selected acoustic feature group (step S64). Next, the symptom prediction unit 125D applies the acoustic feature included in the model construction acoustic feature output in step S64 to a symptom prediction model indicated by the model ID included in the model construction acoustic feature (using the symptom prediction model), and outputs (predicts) a symptom prediction result obtained as a result of the application (step S65).

[0095] In the flowchart shown in FIG. 12, step S61 may be executed at any time after step S60 and before step S64.

[0096] Regarding the first to fourth embodiments described above, the functional blocks and sub-functional blocks of each embodiment can be appropriately applied and combined with other embodiments.

[0097] Next, the effects of the symptom prediction device 1 according to the embodiment will be described.

[0098] The symptom prediction device 1 includes an acquisition unit 11 that acquires voice information related to the user's voice, and a prediction unit 12 that makes a prediction regarding the symptoms of heart failure or renal failure of the user based on the voice information acquired by the acquisition unit 11. With this configuration, it is possible to make a prediction regarding the symptoms of heart failure or renal failure.

[0099] In addition, in the symptom prediction device 1, the prediction unit 12 may perform prediction based on voice change information regarding a voice change obtained based on voice information. With this configuration, for example, it is possible to perform more accurate prediction of symptoms of heart failure or renal failure based on the voice change indicated by the voice change information.

[0100] Furthermore, in the symptom prediction device 1, the prediction unit 12 may perform prediction using a prediction model that performs prediction according to the degree of voice change. With this configuration, for example, it is possible to perform more accurate prediction of symptoms of heart failure or renal failure using a prediction model that performs prediction according to the degree of voice change.

[0101] In addition, in the symptom prediction device 1, the prediction unit 12 may perform prediction further based on past voice change information. With this configuration, it is possible to perform more accurate prediction of symptoms of heart failure or renal failure further based on past voice changes indicated by the past voice change information.

[0102] Furthermore, in the symptom prediction device 1, the prediction unit 12 may perform prediction based on an acoustic feature amount obtained based on voice information. With this configuration, it is possible to perform more accurate prediction of symptoms of heart failure or renal failure based on the acoustic feature amount.

[0103] The symptom prediction device 1 of the present disclosure may have the following configuration.

[0104] [1] An acquisition unit that acquires voice information related to a user's voice; a prediction unit that predicts a symptom of heart failure or renal failure of the user based on the voice information acquired by the acquisition unit; A symptom prediction device comprising:

[0105] [2] The prediction unit performs the prediction based on voice change information regarding a voice change obtained based on the voice information. The symptom prediction device according to [1].

[0106] [3] The prediction unit performs the prediction using a prediction model that performs the prediction according to the degree of the voice change. The symptom prediction device described in [2].

[0107] [4] The prediction unit performs the prediction further based on the past voice change information. The symptom prediction device according to [2] or [3].

[0108] [5] The prediction unit performs the prediction based on an acoustic feature obtained based on the speech information. The symptom prediction device according to any one of [1] to [4].

[0109] The symptom prediction device 1 predicts the worsening of symptoms of heart failure and renal failure based on the voice. The symptom prediction device 1 predicts symptoms of heart failure and renal failure from the voice. In this case, the user's voice may be high or low in hoarseness. Therefore, the symptom prediction device 1 learns two or more types of models corresponding to the hoarseness level. During inference, the symptom prediction device 1 has the user speak and extracts acoustic features to be used for symptom prediction. At the same time, the symptom prediction device 1 extracts at least one of CPP and CSID as acoustic features to be used for hoarseness determination. The symptom prediction device 1 determines the hoarseness level by setting a threshold value for the extracted features. The symptom prediction device 1 selects a symptom prediction model according to the hoarseness level determination result, and inputs the acoustic features for symptom prediction.

[0110] The symptom prediction device 1 is a device that predicts symptoms of heart failure or renal failure from an input voice, and may be a device that has two or more types of symptom prediction models corresponding to the degree of hoarseness, judges the degree of hoarseness based on at least one of CPP or CSID extracted from the input voice, and selects a symptom prediction model depending on the result of the hoarseness degree judgment.

[0111] The symptom prediction device 1 is a device inspired by the fact that when a patient has chronic hoarseness, the tendency of the acoustic feature in the patient's voice differs between when the patient's symptoms are worsening and when the patient has not, and that the tendency of the change in the acoustic feature in the patient's voice differs depending on whether the patient has chronic hoarseness or not. If the patient has hoarseness, it becomes difficult to properly extract the acoustic feature for symptom prediction (the hoarseness is added as noise), so it is considered that the change in the value of the acoustic feature for symptom prediction depending on whether the patient has a chronic hoarseness becomes small for a person with chronic hoarseness. On the other hand, the above effect is not observed in a voice without hoarseness. As a result, the tendency of the change in the feature differs depending on whether the patient has chronic hoarseness or not. According to the symptom prediction device 1, the tendency of the change in the feature extracted in the acoustic feature for symptom prediction extraction unit differs depending on whether the patient has chronic hoarseness or not, so there is a possibility that the feature cannot be picked up well even if it is input to the same model in the symptom prediction model selection unit 123B or the like. Therefore, the symptom prediction model selection unit 123B can select a model formed from voices with similar tendencies according to the degree of hoarseness, thereby improving the estimation accuracy.

[0112] The block diagrams used to explain the above embodiments show functional blocks. These functional blocks (components) are realized by any combination of at least one of hardware and software. The method of realizing each functional block is not particularly limited. That is, each functional block may be realized by using one device that is physically or logically combined, or may be realized by using two or more devices that are physically or logically separated and directly or indirectly connected (for example, by wire, wirelessly, etc.). The functional blocks may be realized by combining the one device or the multiple devices with software.

[0113] Functions include, but are not limited to, judgment, determination, judgment, calculation, computation, processing, derivation, investigation, search, confirmation, reception, transmission, output, access, resolution, selection, selection, establishment, comparison, assumption, expectation, consideration, broadcasting, notifying, communicating, forwarding, configuring, reconfiguring, allocation, mapping, assignment, etc. For example, a functional block (component) that performs the function of transmission is called a transmitting unit or a transmitter. In either case, as described above, there is no particular limitation on the method of realization.

[0114] For example, the symptom prediction device 1 according to an embodiment of the present disclosure may function as a computer that performs processing of the symptom prediction method of the present disclosure. Fig. 13 is a diagram showing an example of a hardware configuration of the symptom prediction device 1 according to an embodiment of the present disclosure. The above-mentioned symptom prediction device 1 may be physically configured as a computer device including a processor 1001, a memory 1002, a storage 1003, a communication device 1004, an input device 1005, an output device 1006, a bus 1007, and the like.

[0115] In the following description, the term "apparatus" can be replaced with a circuit, a device, a unit, etc. The hardware configuration of the symptom prediction device 1 may be configured to include one or more of the devices shown in the figure, or may be configured to exclude some of the devices.

[0116] Each function in the symptom prediction device 1 is realized by loading a specific software (program) onto hardware such as a processor 1001 and a memory 1002, causing the processor 1001 to perform calculations, control communication via a communication device 1004, and control at least one of reading and writing of data in the memory 1002 and the storage 1003.

[0117] The processor 1001, for example, operates an operating system to control the entire computer. The processor 1001 may be configured with a central processing unit (CPU) including an interface with peripheral devices, a control device, an arithmetic unit, a register, etc. For example, the above-mentioned acquisition unit 11 and prediction unit 12, etc. may be realized by the processor 1001.

[0118] Moreover, the processor 1001 reads out a program (program code), a software module, data, etc. from at least one of the storage 1003 and the communication device 1004 into the memory 1002, and executes various processes according to the read out programs. As the program, a program that causes a computer to execute at least a part of the operations described in the above-mentioned embodiment is used. For example, the acquisition unit 11 and the prediction unit 12 may be realized by a control program stored in the memory 1002 and operating in the processor 1001, and other functional blocks may be similarly realized. Although the above-mentioned various processes have been described as being executed by one processor 1001, they may be executed simultaneously or sequentially by two or more processors 1001. The processor 1001 may be implemented by one or more chips. The program may be transmitted from a network via a telecommunication line.

[0119] The memory 1002 is a computer-readable recording medium, and may be configured by at least one of, for example, a Read Only Memory (ROM), an Erasable Programmable ROM (EPROM), an Electrically Erasable Programmable ROM (EEPROM), a Random Access Memory (RAM), etc. The memory 1002 may be called a register, a cache, a main memory (primary storage device), etc. The memory 1002 can store a program (program code), software modules, etc. that are executable to implement a wireless communication method according to an embodiment of the present disclosure.

[0120] Storage 1003 is a computer-readable recording medium, and may be composed of at least one of, for example, an optical disk such as a CD-ROM (Compact Disc ROM), a hard disk drive, a flexible disk, a magneto-optical disk (e.g., a compact disk, a digital versatile disk, a Blu-ray (registered trademark) disk), a smart card, a flash memory (e.g., a card, a stick, a key drive), a floppy (registered trademark) disk, a magnetic strip, and the like. Storage 1003 may be referred to as an auxiliary storage device. The above-mentioned storage medium may be, for example, a database, a server, or other suitable medium including at least one of memory 1002 and storage 1003.

[0121] The communication device 1004 is hardware (transmission / reception device) for performing communication between computers via at least one of a wired network and a wireless network, and is also called, for example, a network device, a network controller, a network card, a communication module, etc. The communication device 1004 may be configured to include a high-frequency switch, a duplexer, a filter, a frequency synthesizer, etc., in order to realize at least one of Frequency Division Duplex (FDD) and Time Division Duplex (TDD). For example, the above-mentioned acquisition unit 11 and prediction unit 12, etc. may be realized by the communication device 1004.

[0122] The input device 1005 is an input device (e.g., a keyboard, a mouse, a microphone, a switch, a button, a sensor, etc.) that accepts input from the outside. The output device 1006 is an output device (e.g., a display, a speaker, an LED lamp, etc.) that performs output to the outside. The input device 1005 and the output device 1006 may be integrated into one device (e.g., a touch panel).

[0123] In addition, each device such as the processor 1001 and the memory 1002 is connected by a bus 1007 for communicating information. The bus 1007 may be configured using a single bus, or may be configured using different buses between each device.

[0124] The symptom prediction device 1 may be configured to include hardware such as a microprocessor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a programmable logic device (PLD), or a field programmable gate array (FPGA), and some or all of the functional blocks may be realized by the hardware. For example, the processor 1001 may be implemented using at least one of these pieces of hardware.

[0125] Notification of information is not limited to the aspects / embodiments described in this disclosure and may be performed using other methods.

[0126] Each aspect / embodiment described in the present disclosure may be applied to at least one of systems using LTE (Long Term Evolution), LTE-A (LTE-Advanced), SUPER 3G, IMT-Advanced, 4G (4th generation mobile communication system), 5G (5th generation mobile communication system), FRA (Future Radio Access), NR (new Radio), W-CDMA (registered trademark), GSM (registered trademark), CDMA2000, UMB (Ultra Mobile Broadband), IEEE 802.11 (Wi-Fi (registered trademark)), IEEE 802.16 (WiMAX (registered trademark)), IEEE 802.20, UWB (Ultra-WideBand), Bluetooth (registered trademark), and other suitable systems, and next-generation systems extended based on these. In addition, a combination of multiple systems (e.g., a combination of at least one of LTE and LTE-A and 5G, etc.) may be applied.

[0127] The order of the steps, sequences, flow charts, etc. of each aspect / embodiment described in this disclosure may be changed unless inconsistent. For example, the methods described in this disclosure present elements of various steps using an example order, and are not limited to the particular order presented.

[0128] The input and output information may be stored in a specific location (e.g., memory) or may be managed using a management table. The input and output information may be overwritten, updated, or added to. The output information may be deleted. The input information may be transmitted to another device.

[0129] The determination may be based on a value represented by a single bit (0 or 1), a Boolean value (true or false), or a numerical comparison (e.g., comparison with a predetermined value).

[0130] Each aspect / embodiment described in this disclosure may be used alone, in combination, or switched according to execution. In addition, notification of specific information (e.g., notification that "X is the case") is not limited to being done explicitly, but may be done implicitly (e.g., not notifying the specific information).

[0131] Although the present disclosure has been described in detail above, it is clear to those skilled in the art that the present disclosure is not limited to the embodiments described in the present disclosure. The present disclosure can be implemented in modified and altered forms without departing from the spirit and scope of the present disclosure as defined by the claims. Therefore, the description of the present disclosure is intended to be illustrative and does not have any limiting meaning on the present disclosure.

[0132] Software shall be construed broadly to mean instructions, instruction sets, code, code segments, program code, programs, subprograms, software modules, applications, software applications, software packages, routines, subroutines, objects, executable files, threads of execution, procedures, functions, etc., whether referred to as software, firmware, middleware, microcode, hardware description language, or otherwise.

[0133] Additionally, software, instructions, information, etc. may be transmitted or received over a transmission medium. For example, if the software is transmitted from a website, server, or other remote source using wired and / or wireless technologies (such as coaxial cable, fiber optic cable, twisted pair, Digital Subscriber Line (DSL)), and / or wireless technologies (such as infrared, microwave, etc.), then these wired and / or wireless technologies are included within the definition of transmission media.

[0134] The information, signals, etc. described in this disclosure may be represented using any of a variety of different technologies. For example, the data, instructions, commands, information, signals, bits, symbols, chips, etc. that may be referred to throughout the above description may be represented by voltages, currents, electromagnetic waves, magnetic fields or magnetic particles, optical fields or photons, or any combination thereof.

[0135] In addition, terms explained in this disclosure and terms necessary for understanding this disclosure may be replaced with terms having the same or similar meanings.

[0136] As used in this disclosure, the terms "system" and "network" are used interchangeably.

[0137] Furthermore, the information, parameters, etc. described in this disclosure may be expressed using absolute values, may be expressed using relative values ​​from a predetermined value, or may be expressed using other corresponding information.

[0138] The names used for the above parameters are not limiting in any way, and furthermore, the formulas etc. using these parameters may differ from those explicitly disclosed in this disclosure.

[0139] As used in this disclosure, the terms "determining" and "determining" may encompass a wide variety of actions. "Determining" and "determining" may include, for example, judging, calculating, computing, processing, deriving, investigating, looking up, search, inquiry (e.g., searching in a table, database, or other data structure), ascertaining, and the like. "Determining" and "determining" may also include receiving (e.g., receiving information), transmitting (e.g., sending information), input, output, accessing (e.g., accessing data in a memory), and the like. In addition, "judgment" and "decision" can include considering resolving, selecting, choosing, establishing, comparing, etc., to be a "judgment" or "decision." In other words, "judgment" and "decision" can include considering some action to be a "judgment" or "decision." In addition, "judgment" can be interpreted as "assuming," "expecting," "considering," etc.

[0140] The terms "connected," "coupled," or any variation thereof, refer to any direct or indirect connection or coupling between two or more elements, and may include the presence of one or more intermediate elements between two elements that are "connected" or "coupled" to each other. The coupling or connection between elements may be physical, logical, or a combination thereof. For example, "connected" may be read as "access." As used in this disclosure, two elements may be considered to be "connected" or "coupled" to each other using at least one of one or more wires, cables, and printed electrical connections, as well as electromagnetic energy having wavelengths in the radio frequency range, microwave range, and optical (both visible and invisible) range, as some non-limiting and non-exhaustive examples.

[0141] As used in this disclosure, the phrase "based on" does not mean "based only on," unless expressly stated otherwise. In other words, the phrase "based on" means both "based only on" and "based at least on."

[0142] Any reference to an element using a designation such as "first," "second," etc., used in this disclosure does not generally limit the quantity or order of those elements. These designations may be used in this disclosure as a convenient method of distinguishing between two or more elements. Thus, a reference to a first and a second element does not imply that only two elements may be employed or that the first element must precede the second element in some way.

[0143] The "means" in the configuration of each of the above devices may be replaced with "part," "circuit," "device," etc.

[0144] When the terms "include," "including," and variations thereof are used in this disclosure, these terms are intended to be inclusive, similar to the term "comprising." Further, when used in this disclosure, the term "or" is not intended to be an exclusive or.

[0145] In this disclosure, where articles have been added due to translation, such as, for example, a, an, and the in English, the disclosure may include that the nouns following these articles are in the plural form.

[0146] In the present disclosure, the term "A and B are different" may mean "A and B are different from each other." The term may also mean "A and B are each different from C." Terms such as "separate" and "coupled" may also be interpreted in the same way as "different." [Explanation of symbols]

[0147] 1...symptom prediction device, 10...storage unit, 11...acquisition unit, 12, 12A, 12B, 12C, 12D...prediction unit, 120A, 120B, 120C, 120D...acoustic feature extraction unit for symptom prediction, 121A, 121B, 121C, 121D...acoustic feature extraction unit for hoarseness determination, 122B, 122C, 122D...hoarseness degree determination unit, 123B, 123C...symptom prediction model selection unit, 124D...feature selection unit, 125A, 125B, 125C, 125D...symptom prediction unit, 1001...processor, 1002...memory, 1003...storage, 1004...communication device, 1005...input device, 1006...output device, 1007...bus.

Claims

1. An acquisition unit that acquires voice information related to a user's voice; a prediction unit that predicts a symptom of heart failure or renal failure of the user based on the voice information acquired by the acquisition unit; A symptom prediction device comprising:

2. The prediction unit performs the prediction based on voice change information regarding a voice change obtained based on the voice information. The symptom prediction device according to claim 1 .

3. The prediction unit performs the prediction using a prediction model that performs the prediction according to the degree of the voice change. The symptom prediction device according to claim 2 .

4. The prediction unit performs the prediction further based on the past voice change information. The symptom prediction device according to claim 2 .

5. The prediction unit performs the prediction based on an acoustic feature obtained based on the speech information. The symptom prediction device according to any one of claims 1 to 4.

6. An acquisition step of acquiring speech information related to a user's speech; a prediction step of predicting a symptom of heart failure or renal failure of the user based on the voice information acquired in the acquisition step; A symptom prediction method comprising:

Images