Dental treatment pain prediction

Abstract

A method and system for generating predictive information including a predictive metric of treatment pain associated with each step of a series of steps to be performed by a dental practitioner as part of a dental treatment. The information may be output as guidance for the practitioner, allowing them to adjust their therapeutic style or method for each step to preemptively relieve the expected pain.

Description

Technical Field

[0001] The present invention relates to the field of oral healthcare, particularly in a clinical setting. Background Art

[0002] Anticipated treatment discomfort or pain is a significant barrier to patients attending dental practitioners regularly and receiving treatment compliance. Treatment pain can contribute to dental anxiety in patients. The experience of pain during previous dental visits has been shown to be associated with dental anxiety. An estimated 3–16% of adults suffer from dental phobia. Patient studies have shown that patients believe that dental practitioners (DPs) should be aware of any discomfort (e.g., pain, stress, or anxiety) during treatment and that the DP should take steps to improve the situation. However, anticipated treatment discomfort is not a major cause of anxiety in the patient population as a whole. In studies with samples representative of the general population aged 15 years and older, 60% reported having at least one dental experience involving “intense pain,” with only 5–6% experiencing very painful dental treatments overall.

[0003] In addition to affecting patients, treating pain and anxiety are also issues that affect dental practitioners. The survey found that 60% of DPs believe that dentistry is more stressful than other professions. The top 5 most significant stressors are "causing pain" (second place) and "anxious patients" (fifth place).

[0004] Therefore, if dental treatment steps can be applied in a way that minimizes pain for the patient, thereby reducing anxiety about future treatment, there is a mutually beneficial outcome.

[0005] Dental practitioners often try to do this as part of their standard practice by anticipating steps that may cause more pain than others and giving clear warnings and reassurances to the patient, as well as performing additional care or work more slowly during these steps. For example, particularly more experienced DPs are able to perform a single treatment with a number of different treatment styles or philosophies, and are able to change treatment styles depending on the pain the patient experiences at each step, e.g., speeding up treatment steps that cause less pain and taking a careful approach when pain is anticipated.

[0006] However, it is difficult in practice to accurately estimate the amount of pain associated with a specific step of treatment and the specific style applied to that step. The pain experienced can also vary depending on the patient's pain threshold. Dentistry physically limits the patient's ability to communicate verbally during the treatment phase of the consultation. These ineffective communications can lead to misunderstandings about the pain the patient may be experiencing. Many dental practitioners will ask patients to raise their hands when they experience (too much) pain. However, by the time the patient raises their hands, he / she is already in pain. Furthermore, there may be a delay of several seconds between the patient indicating pain and the dental practitioner adjusting their treatment style to improve the pain. Additionally, the DP often has no way of knowing whether the patient is comfortable and, therefore, whether a specific treatment step may become more intense. Summary of the invention

[0007] In view of the above, the inventors recognized that it would be beneficial to provide dental practitioners (DPs) with a means to more clearly determine the expected treatment pain for a given patient during each step of a particular treatment. These insights into treatment pain can help guide dental practitioners to adapt their style or approach to each step to alleviate the expected pain.

[0008] Furthermore, the inventors have realised that it would be beneficial for the DP to be able to estimate the expected treatment discomfort for specific steps in the procedure, taking into account also the dentist's treatment style (i.e., the DP's approach or attitude towards performing certain treatments). Using this knowledge, the DP can, for example, expedite treatment steps that cause less pain, while taking a more careful approach when pain is expected. In this way, dentists can adapt their treatment style to the needs of their patients.

[0009] The invention is defined by the claims.

[0010] According to an example according to one aspect of the present invention, a computer-implemented method is provided for providing guidance on dental treatment to a dental practitioner. The method includes running a prediction algorithm. The prediction algorithm is configured to receive input variables, the input variables including at least: a type of treatment required by a patient, for example, selected from a predefined set of possible treatment types. The prediction algorithm is also configured to generate an output prediction data set, the output prediction data set including a prediction measure of treatment pain for each step in a series of steps included in the type of input treatment.

[0011] In some embodiments, the method further includes running a guidance algorithm configured to: access a predictive data set; generate guidance information indicating predicted treatment pain for at least a subset of a series of treatment steps; and control a user interface to generate at least one user-perceivable output indicating the guidance information.

[0012] In some embodiments, the input variables that the predictive algorithm is configured to take as input may also include the type of treatment to be performed by the dental practitioner, for example selected from a predefined set of possible treatment types.

[0013] It is therefore proposed to use an algorithm to predict in advance the pain associated with a particular step of a particular treatment, and optionally given the dental practitioner's treatment style. This can then preferably be used to generate guidance for the dental practitioner. This can be generated in real time during treatment, or can be generated in advance, for example in a report, which the dental practitioner can then use to plan their approach to the different steps of the treatment to preemptively try to improve or reduce pain in the different steps.

[0014] In some embodiments, the prediction algorithm includes at least one trained machine learning algorithm.

[0015] As an example, a machine learning algorithm can be trained using a training database that includes real-world measured data about pain associated with different steps of treatment, and optionally for different treatment styles. In this way, the algorithm can be trained to classify the expected treatment pain level or amount for a particular treatment step and optionally a given treatment style.

[0016] An embodiment provides an improved patient experience due to better management of treatment pain. It provides an overall improved health outcome for the patient by predicting pain during treatment and providing DP guidance to manage pain. This can improve patient compliance with visits to treatment providers. Another advantage is improved staff experience by reducing occupational stress for dentists. Another advantage is reduced cost of care by reducing visit time.

[0017] Regarding treatment styles, these treatment styles are defined according to a predefined set of possible treatment styles. Treatment types are treatment type classifications, i.e., a set of classifications indicating the type of administration of a treatment. The predefined set of possible treatment styles can be defined by a list or a data record. Treatment styles can be custom defined or can be defined by specific clinical criteria.

[0018] As an example, a predefined set of possible treatment styles may include one or more of the following: a slow, meticulous approach; a faster, forceful approach; a mechanical approach; a pharmaceutical approach. Another example is a minimal anesthesia approach, in which anesthesia use is minimized. Another approach is a preventive antibiotic approach, in which antibiotics are administered on a "just in case" basis after any dental treatment. Another approach is a minimal antibiotic approach, in which antibiotics are administered only when necessary. Another approach is a fast / flexible approach, in which the practitioner will be given the option of performing the required treatment directly when performing the dental examination (even though this may be more painful than waiting for an appointment for a visit in about a few weeks, as the latter can be prepared for anesthesia). Another approach is a preventive treatment approach, in which superficial caries are treated with mechanical (drilling) treatment. Another approach (the opposite of a preventive treatment approach) is a minimal treatment approach, in which the response to superficial caries is to prescribe a high-fluoride toothpaste, and / or give a fluoride rinse, and schedule a monitoring appointment.

[0019] In some embodiments, the method may include controlling the user interface to generate an ordered sequence of user-perceivable outputs, each user-perceivable output indicating predicted treatment pain for a treatment step in a series of treatment steps. Thus, guidance may be generated sequentially or synchronously with the treatment steps, providing guidance to the dental practitioner prior to each step of the treatment.

[0020] The generation of each next user-perceivable output may be triggered by a user input from a user interface, or for example according to a predefined timing schedule. The timing schedule may be predefined before the guiding algorithm is executed based on the user input.

[0021] In some embodiments, the method includes accessing a scheduling database storing records indicating scheduled treatment sessions, each record including at least an indication of a type of treatment to be performed. The method may also include receiving information indicating a treatment style to be performed by the dental practitioner, and optionally wherein the treatment style to be performed is also included as part of each data record in the scheduling database. The method may also include generating input variables for a prediction algorithm based on said accessing step and said receiving step.

[0022] In some embodiments, the input variables to the prediction algorithm also include one or more patient oral health characteristics.

[0023] In some embodiments, the predictive algorithm comprises a machine learning algorithm pre-trained using a training database comprising corresponding data entries for each of multiple previous visits by multiple different patients to a set of different dental treatment providers, and wherein each visit entry comprises at least an indication of: the type of treatment administered; and a measure indicating treatment pain for each step in a series of steps comprised in the treatment.

[0024] In some embodiments, each visit entry in the training database further includes an indication of a treatment style, wherein the treatment style is one of the predefined set of treatment styles.

[0025] In some embodiments, at least a subset of visit entries may also include an indication of one or more of: duration of treatment; cost of treatment; one or more oral health characteristics of the patient; geographic location of the treating provider; and / or one or more patient personality or demographic characteristics.

[0026] In some embodiments, geographic location may be used, for example, to benchmark dental practitioners relative to practitioners in similar geographic areas with respect to pain experience with different treatments.

[0027] In some embodiments, the predictive metric indicative of treatment pain for each step in a series of steps of treatment is a quantitative metric.

[0028] In some embodiments, within each visit entry in the data set, for each of one or more pain response episodes for each treatment step, a metric indicating treatment pain for each step in a series of steps corresponds to a total measured amount of pain corresponding to the product of the measured pain response level and the time the pain response level was experienced.

[0029] In some embodiments, within each visit entry, the metric indicative of treatment pain for each step in a series of steps includes the maximum or peak pain level experienced during the corresponding treatment step.

[0030] In some embodiments, within each visit entry, a measure of treatment pain for each step in a series of steps includes a measure of the maximum rate of change in pain level experienced during each treatment step. A rapid increase in pain level (e.g., from a zero pain level or a low pain level) can give a significant pain sensation even if the pain level reached is not high.

[0031] In some embodiments, the method further comprises obtaining a metric indicative of estimated actual treatment pain for at least a subset of the treatment steps.

[0032] In some embodiments, the method further comprises comparing the estimated actual treatment pain to the predicted treatment pain for each step in at least a subset of the steps.

[0033] In some embodiments, the method further comprises generating feedback information to the dental practitioner based on the comparison, eg, indicating a result of the comparison.

[0034] By comparing the estimated treatment pain to the actual treatment pain, this allows the dental practitioner to take into account any systematic differences between prediction and reality. In some embodiments, the method may include adjusting the prediction metric for treatment pain based on differences between predicted and actual treatment pain detected for at least a subset of previous steps in the treatment.

[0035] In some embodiments, during performance of each step in a subset of treatment steps by a dental practitioner, a metric indicative of an estimated actual treatment pain for each treatment step in at least one subset of treatment steps is received in real time, and wherein comparison and feedback generation is performed in real time upon receipt of the estimated actual treatment pain for each step.

[0036] This allows dentists to tailor their approach to optimize the balance between pain and, for example, speed.

[0037] In some embodiments, obtaining a metric indicative of estimated actual treatment pain for at least a subset of treatment steps comprises:

[0038] receiving, during each step of at least a subset of the treatment steps, a biosignal of the subject from a biosignal sensing device coupled to the subject;

[0039] Identify physiological response events in biological signals;

[0040] A metric indicative of actual treatment pain estimated for each of the steps of the treatment is determined based on the physiological response events identified for that step.

[0041] This provides a technical means for measuring the actual pain associated with each treatment set. This will be explained in further detail below.

[0042] The biological signal may be a physiological parameter signal. In some embodiments, the biological signal may be one of the following: a heart rate signal and a skin conduction sensor signal.

[0043] In some embodiments, obtaining a metric indicative of estimated actual treatment pain for at least a subset of the treatment steps may further include:

[0044] receiving an indication of the timing of a series of oral contact events, each contact event corresponding to mechanical engagement of an oral implement with a surface within an oral cavity of a subject; and

[0045] A temporal proximity between each physiological response event and an immediately preceding contact event is determined, and based thereon, a classification of the subject's pain and / or non-pain arousal associated with the physiological response event is performed.

[0046] In some embodiments, the method further comprises: receiving data indicating motion and / or force patterns of the dental tool during at least one step of the treatment; and applying a treatment style analyzer algorithm configured to estimate a treatment style associated therewith based on the motion and / or force patterns.

[0047] The treatment style analyzer algorithm may be a machine learning algorithm that is trained to generate a treatment style classification based on input motion and / or force data.

[0048] In some embodiments, the method may further include re-running the prediction algorithm using the treatment style estimated by the treatment style analyzer algorithm to generate a new prediction of the measure of treatment pain for at least one step. In some embodiments, the method may further include generating a user-perceivable output indicating the new prediction.

[0049] For example, this can be applied in real time during treatment to measure the treatment style in real time (e.g., using pressure and motion sensors in the treatment tool). Thus, the estimated pain level per step can be updated in real time, for example, based on the treatment style.

[0050] Another aspect of the invention is a computer program product comprising code means configured to, when run on a processor, cause the processor to perform a method according to any embodiment set forth in this document or according to any claim of this application.

[0051] Another aspect of the present invention is a processing unit, comprising: an input / output; and one or more processors. The one or more processors are adapted to run a prediction algorithm. The prediction algorithm may optionally include at least one trained machine learning algorithm. The prediction algorithm is configured to: receive input variables, which include at least: a type of treatment required by a patient, and optionally a type of treatment performed by a dental practitioner, each treatment type selected from a predefined set of possible treatment types and styles, and generate an output prediction data set, the output prediction data set including a prediction measure of treatment pain for each step in a series of steps included in the input treatment type.

[0052] In some embodiments, the one or more processors may be further adapted to run a guidance algorithm. The guidance algorithm may be configured to: access the prediction data set, generate guidance information indicating predicted treatment pain for at least a subset of the series of treatment steps, and control the user interface to generate at least one user-perceivable output indicating the guidance information.

[0053] Any features or embodiments described with respect to the method aspects of the invention may also be applied or incorporated into the processing unit aspects.

[0054] These and other aspects of the invention will be apparent from and elucidated with reference to the embodiments described hereinafter. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] For a better understanding of the invention, and in order to more clearly show how the invention may be carried into effect, reference will now be made, by way of example only, to the accompanying drawings, in which:

[0056] Figure 1 The steps of an example method according to one or more embodiments of the present invention are outlined;

[0057] Figure 2 Schematically outlines components of an example processing device and a processing flow according to one or more embodiments of the present invention;

[0058] Figure 3 schematically illustrates a process flow according to one or more embodiments of the present invention; and

[0059] Figure 4 The processing flow according to another one or more embodiments of the present invention is schematically shown. Specific implementation style

[0061] The present invention will be described with reference to the accompanying drawings.

[0062] It should be understood that the detailed description and specific examples, although indicating exemplary embodiments of the apparatus, system and method, are for illustrative purposes only and are not intended to limit the scope of the invention. These and other features, aspects and advantages of the apparatus, system and method of the present invention will become better understood from the following description, the appended claims and the accompanying drawings. It should be understood that the drawings are only schematic and are not drawn to scale. It should also be understood that the same reference numerals are used throughout the drawings to represent the same or similar components.

[0063] The present invention provides a method and system for generating predictive information, including a predicted measure of treatment pain associated with each step in a series of steps that will be performed by a dental practitioner as part of a dental treatment. For example, a machine learning algorithm can be used that is trained on historical pain measurement information for the same type of treatment. This allows the algorithm to learn from historical pain associated with different steps. This information can be output as a guide for the practitioner, allowing them to adjust their treatment style or approach for each step to preemptively reduce the expected pain.

[0064] Figure 1 The steps of an example method according to one or more embodiments are summarized in block diagram form. These steps will be generally described before further explanation in the form of an example embodiment.

[0065] Method 8 is a computer-implemented method for providing guidance to a dental practitioner regarding dental treatment.

[0066] The method 8 includes running a prediction algorithm 10. The prediction algorithm is configured to receive 12 input variables, the input variables including at least: a type of treatment desired by the patient, for example selected from a predefined set of possible treatment types. The prediction algorithm is also configured to generate 14 an output prediction data set, the output prediction data set including a prediction measure of treatment pain for each step in a series of steps included in the type of input treatment.

[0067] The input received by the predictive algorithm may also include a treatment style implemented by a dental practitioner, for example selected from a predefined set of possible treatment styles.

[0068] In some embodiments, the method 8 may further include running a guidance algorithm, guidance algorithm 20. The guidance algorithm may be configured to: access 22 the prediction data set; generate 24 guidance information indicating predicted treatment pain for at least a subset of the series of treatment steps; and control 26 a user interface to generate at least one user-perceivable output indicating the guidance information.

[0069] In some embodiments, the prediction algorithm includes at least one trained machine learning algorithm.

[0070] The method may for example comprise controlling 26 a user interface to generate an ordered sequence of user-perceivable outputs, each user-perceivable output indicating a predicted treatment pain for a treatment step in a series of treatment steps. The generation of each next user-perceivable output may be triggered by a user input from the user interface, or according to a predefined timing schedule. The timing schedule may be predefined prior to executing the guidance algorithm based on the user input.

[0071] In some embodiments, the input variables to the prediction algorithm also include one or more patient oral health characteristics.

[0072] In this case, the prediction algorithm can be pre-trained to generate predicted pain tailored to a specific health characteristic or characteristics. For example, an illustrative example is whether the patient is prone to gingivitis, in which case more pain can be predicted. Other examples of oral health characteristics can include, as illustrations: the presence or absence of sensitive gums, the presence of infection, a recent history of intervention (e.g., tooth extraction, periodontal pocket treatment), the presence of root abscesses or injuries, or jaw problems (e.g., TMJ-jaw opening pain).

[0073] In some advantageous embodiments, the method may further include accessing a scheduling database storing records indicating scheduled therapy sessions, each record including at least an indication of a type of therapy to be performed. This step results in obtaining information about the type of therapy administered to a particular patient. For example, the scheduling database may be queried using the patient identification information to obtain a scheduling record for the relevant patient and thereby obtain an indication of the type of therapy.

[0074] Optionally, the method may further comprise receiving information indicating a treatment style to be implemented by the dental practitioner. For example, this may be input by the dental practitioner into a user input device, or the treatment style to be implemented may also be included as part of each data record in the scheduling database and obtained from the record.

[0075] The method may include generating input variables for a prediction algorithm based on the accessing step and the receiving step.

[0076] This thus automates the process of obtaining the required inputs for the prediction algorithm.

[0077] Although in the above description, the various steps of the method are specified as being performed by a specific algorithm, it should be understood that in other examples, the steps of the method can all be performed by a single algorithm, or by more than two algorithms. The concept of the present invention is not limited by any specific division of steps between specific algorithms. This designation is for convenience of description and reflects the most likely implementation in practice.

[0078] Reference has been made to the option of a prediction algorithm that takes into account the treatment style when generating a prediction of the pain measure for each treatment step. For example, the treatment style may include one or more of the following: a slow, meticulous approach; a faster, forceful approach; a mechanical approach; a pharmaceutical approach. Another example is a minimal anesthesia approach, in which anesthesia use is minimized. Another approach is a preventive antibiotic approach, in which antibiotics are administered on a "just in case" basis after any dental treatment. Another approach is a minimal antibiotic approach, in which antibiotics are administered only when necessary. Another approach is a fast / flexible approach, in which the practitioner will be given the option of proceeding directly to the desired treatment when performing a dental examination (even though this may be more painful than waiting for an appointment for a visit in about a few weeks, because the latter can be prepared for anesthesia). Another approach is a preventive treatment approach, in which superficial caries are treated with mechanical (drilling) treatment. Another approach (as opposed to a preventive treatment approach) is a minimal treatment approach, in which the response to superficial caries is to prescribe a high-fluoride toothpaste, and / or give a fluoride rinse, and schedule a monitoring appointment.

[0079] With respect to the method of the present invention, the set of predefined treatment styles may be defined in any desired manner. Each treatment style will be given a label or name, and its definition (in terms of its characteristics) will be predefined. This will be known to the dental practitioner. Then, when providing treatment style information for input to a predictive algorithm, the dental practitioner can use the known definitions to decide what style in the list best matches his or her own style.

[0080] Similarly, when compiling a training database for the optional machine learning algorithm used to train the predictive algorithm, the treatment providers contributing data to the database will also be provided with a list of pre-defined treatment styles with definitions of each of their styles, and in this way they can label or tag each treatment session data entry with the corresponding style that best matches the style used for that treatment.

[0081] Based on the above, with respect to the definition of treatment style, there is consistency in the definition across all parts of the method and system including all databases and algorithms.

[0082] As mentioned above, the method may also be implemented in hardware form, for example in the form of a processing unit configured to perform a method according to any example or embodiment described in this document or according to any claim of this application.

[0083] To further aid understanding, Figure 2 A schematic representation of an example processing unit 32 configured to perform a method according to one or more embodiments of the present invention is shown, and a processing flow is also schematically shown in more detail. The processing unit 32 includes one or more processors (not shown) and may include input / output. The one or more processors are suitable for running a prediction algorithm 34, and the prediction algorithm is configured to: receive input variables, the input variables at least including: the type of treatment 38 required by the patient, for example, selected from a predefined set of possible treatment types. Optionally, the input variables may also include at least a treatment style 40 to be implemented by a dental practitioner. The prediction algorithm 34 is also configured to generate an output prediction data set for each step in a series of steps included in the type of input treatment, the output prediction data set including a prediction metric for treatment pain 42. The prediction algorithm may include at least one trained machine learning algorithm.

[0084] In some embodiments, one or more processors of the processing unit 32 may also be adapted to run a guidance algorithm 20 that is configured to access a predictive data set and generate guidance information 52 indicating predicted treatment pain for at least a subset of a series of treatment steps, and control a user interface 62 to generate at least one user-perceivable output 64 indicating the guidance information.

[0085] For example, the user interface 62 may include a display device for generating a visual user-perceivable output. For example, this may be a display device connected to a computing device located in, for example, a treatment room used by a dental practitioner. In this way, guidance information may be provided to the dental practitioner during treatment. Alternatively, the user interface may be located elsewhere, for example to enable the dental practitioner to view guidance prior to treatment.

[0086] The processing unit 32 can be part of a dental practitioner's computing device, such as a mobile computing device, such as a smart phone or tablet. In other examples, it can be part of a dental device. In some examples, it can be part of a cloud-based server. In some examples, it can be part of a computing workstation.

[0087] By way of illustration, an example implementation of the method according to a set of specific embodiments that the inventors believe to be particularly advantageous can be summarized as including the following features:

[0088] A processor configured to execute an algorithm programmed to:

[0089] receiving information about an upcoming consultation, including treatment type and treatment style / philosophy, and optionally patient characteristics;

[0090] Predict the expected treatment pain for each treatment step; and

[0091] Guidance or training is provided to dental practitioners by generating feedback indicating the expected level or amount of pain for each treatment step during treatment.

[0092] As described above, part of the method may include using a prediction algorithm in the form of a trained machine learning algorithm. The details regarding the trained machine learning algorithm will now be discussed in more detail.

[0093] A machine learning algorithm is any self-training algorithm that treats input data in order to produce or predict output data. In the present case, the input data comprises at least the type of treatment required by the patient and optionally also the style of treatment delivered by the dental practitioner, and the output data comprises a predicted measure of treatment pain for each of a series of steps comprised by the type of input treatment.

[0094] Suitable machine learning algorithms used in the present invention will be apparent to those skilled in the art. Examples of suitable machine learning algorithms include decision tree algorithms and artificial neural networks. Other machine learning algorithms such as logistic regression, support vector machines, natural Bayesian models, and k-nearest neighbor models are suitable alternatives.

[0095] The structure of an artificial neural network (or simply, a neural network) is inspired by the human brain. A neural network includes multiple layers, each layer includes multiple neurons. Each neuron includes a mathematical operation. In particular, each neuron may include a different weighted combination of a single type of transformation (e.g., the same type of transformation with different weights, a sigmoid transformation, etc.). In the process of treating input data, the mathematical operation of each neuron is performed on the input data to produce a digital output, and the output of each layer in the neural network is sequentially fed to the next layer. The last layer provides the output.

[0096] Methods for training machine learning algorithms are well known. Typically, such methods include obtaining a training data set that includes training input data entries and corresponding training output data entries. An initialized machine learning algorithm is applied to each input data entry to generate a predicted output data entry. The error between the predicted output data entry and the corresponding training output data entry is used to modify the machine learning algorithm. The process can be repeated until the error converges and the predicted output data entry is sufficiently similar to the training output data entry (e.g., ±1%). This is generally referred to as a supervised learning technique.

[0097] For example, in the case where the machine learning algorithm is formed by a neural network, the mathematical operation (weight) of each neuron can be modified until the error converges. Known methods for modifying neural networks include gradient descent, back propagation algorithms, etc.

[0098] In the present case, the training data set may be based on data contained in a database comprising a respective data entry for each of a plurality of previous / historical treatment sessions by a plurality of different patients with a set of different dental treatment providers, and wherein each treatment session entry comprises at least an indication of:

[0099] the type of treatment administered;

[0100] A measure of pain for each of a series of steps included in the treatment is indicated and preferably also includes an indication of:

[0101] A treatment style, wherein the treatment style is one of the predefined set of treatment styles.

[0102] In this case, the training input data entries of the training data may correspond to the type of treatment required by the patient and optionally to the treatment style. The training output data entries may correspond to a measure of predicted pain for each step in a series of steps indicative of the type of treatment.

[0103] Another alternative is to use a prediction algorithm comprising a plurality of trained machine learning algorithms, wherein each trained machine learning algorithm has been trained using a filtered subset of the training data set, the filtered subset comprising only those training data entries corresponding to one of the particular respective treatment styles. In this case, the training input data for each machine learning algorithm may comprise only (at least) the treatment type, and the training output data entries may comprise only (at least) a predicted measure of treatment pain. Then, when the prediction algorithm is run in the context of the present invention, the algorithm may use an input variable indicative of the treatment style implemented by the dental practitioner to select a particular one of the plurality of trained machine learning algorithms that has been trained to generate pain predictions for a particular input treatment style. The machine learning algorithm may then be run by the prediction algorithm having an input comprising the type of treatment desired by the patient, and wherein the algorithm generates an output indicative of a predicted pain measure for each step in a series of steps of treatment for the particular treatment style indicated by the input variable.

[0104] In some embodiments, at least a subset of the treatment session data entries in the training database may also include an indication of one or more of: duration of treatment; cost of treatment; one or more oral health characteristics of the patient; geographic location of the treatment provider; and / or one or more patient personality characteristics; and wherein the machine learning algorithm is trained accordingly.

[0105] In some embodiments, the metric indicating the pain of treatment for each step in the series of steps included in the data entry is a quantitative metric. For example, it may correspond to the total amount of pain measured. For example, for each of one or more pain response episodes, the total measured amount of pain may correspond to the product of the measured pain response level and the time the pain response level is experienced. Alternatively, the metric indicating the pain of treatment for each step in the series of steps may correspond to the maximum or peak pain level experienced during the corresponding treatment step. In further embodiments, the metric indicating the pain of treatment for each step in the series of steps may correspond to the rate of change of the pain level experienced during the corresponding treatment step, or the maximum rate of change of the pain level experienced during the treatment step.

[0106] For example, pain information may be derived from biosensing signals acquired as a function of time during treatment over multiple historical treatment sessions.

[0107] As an example, during a (historical) treatment session, an objective measure of treatment pain may be derived in the following way for generating data entries for the database.

[0108] A processor at a treatment provider's office may be adapted to classify a subject's response to a series of contact events during a measurement session, the processor being adapted to: receive an indication of the timing of a series of oral contact events during the measurement session, each contact event corresponding to a mechanical engagement of an oral tool with a surface within the subject's oral cavity; receive a biosignal of the subject from a biosignal sensing device coupled to the subject during the measurement session; identify a physiological response event in the biosignal; determine a temporal proximity between each physiological response event and an immediately preceding contact event, and based thereon classify the subject's pain and / or non-pain arousal associated with the physiological response event. The physiological response event may take the form of a fluctuation or oscillation in the signal, such as a peak or dip in the signal. For example, the processor may be configured to classify a physiological response event as being associated with a pain state based on the physiological response event occurring within a predefined time window immediately following the contact event. The classification may also include determining a pain level of the pain state based on one or more characteristics of a peak or dip of the physiological response event, and preferably wherein the one or more characteristics include a maximum height of a leading edge of the peak or dip. The classification may be performed in real time during the measurement session for each individual period corresponding to each respective step of the treatment.

[0109] The processor may be adapted to calculate a total amount of pain over a time period spanned by each respective treatment step, the total amount of pain being the product of a pain level and a time during which pain is experienced, the pain level being calculated for each physiological response event peak or dip having a pain classification based on one or more characteristics of a physiological response event peak or dip. The processor may also be configured to classify a physiological response event as being associated with a non-pain arousal state based on a physiological response event occurring outside a time window of a predefined duration following any contact event. The processor may optionally be adapted to further calculate a total amount of non-pain arousal over a time period spanned by each respective treatment step, the total amount of non-pain arousal being the product of arousal level and a time during which arousal is experienced, the arousal level being calculated for each physiological response event peak or dip having a non-pain arousal classification based on one or more characteristics of a physiological response event peak or dip.

[0110] Within a treatment provider's office, there may be a system comprising: a contact identification device for identifying the timing of a series of oral contact events, each contact event corresponding to mechanical engagement of an oral tool with a surface within the subject's oral cavity; a biosignal sensing device (such as a skin conduction sensor or a heart rate sensor or any other biosignal sensor) for measuring a physiological response of the subject; and a processor as already described, arranged to receive, over a measurement session, one or more contact signals from the contact identification device indicating the timing of the series of oral contact events and a biosignal from the biosignal sensing device.

[0111] Thus, it can be appreciated how to obtain an objective measure of pain associated with each step of a particular treatment based on biosignals. For each database treatment session entry, the measure of treatment pain recorded in the database used to train the machine learning algorithm corresponds to the total amount of pain measured on each individual treatment step, or may correspond to the highest pain level experienced during each treatment step, or may correspond to the average pain level over the entire time period covered by each treatment step, or the ratio of total pain to total non-pain amount during each treatment step, or any other variable or parameter derived from the above-mentioned biosignal pain data obtained during each historical treatment session and for each treatment step.

[0112] This information may then be stored as an entry in a database used to train the prediction algorithm 34, along with an indication of the type of treatment administered and optionally the treatment style, where the treatment style is one of the predefined set of treatment styles.

[0113] Indeed, another aspect of the present invention may be to compile the above-mentioned training data set using the above-mentioned pain measurement technique or another objective pain measurement technique.

[0114] Another aspect of the invention may be the training of a prediction algorithm according to the example methods already described above.

[0115] One aspect of the invention may include a combination of the compilation of a training database using the above-described pain measurement techniques and the training of a prediction algorithm using the training database according to the above-described example methods.

[0116] An embodiment of the present invention may include:

[0117] compilation of a training database using, for example, one or more features of the pain measurement technique described above or another technique;

[0118] training of a machine learning algorithm included in a prediction algorithm using the training database; and

[0119] Referenced above Figure 1 Execution of the stated pain prediction method steps and optional guidance generation method steps (or any other embodiment / example of the method outlined in this document, or according to any claim of this application).

[0120] Now it will pass Figure 1 The description of the concept of the present invention summarized in the above is used to describe an example implementation of the method according to a set of specific embodiments of the present invention. It should be understood that not all features of the set of specific embodiments are necessary for the concept of the present invention, and the description of these features is used to help understand and provide an example to illustrate the concept of the present invention.

[0121] According to this set of specific embodiments, a system and method are provided that are configured for providing feedback or training information to a dental practitioner during the treatment portion of a patient consultation. As already discussed, in order to predict expected treatment pain, an algorithm is trained using, for example, a large patient database derived from objective pain measurements recorded using the methods already discussed, based on biosignals measured during historical treatment sessions.

[0122] The training database can, for example, include at least the following information for each case: type of treatment (e.g., cleaning), treatment style or philosophy (e.g., mechanical, meticulous approach, vigorous approach), a measure of treatment pain experienced, preferably during each step of treatment (e.g., a visual analog scale, or skin response), and optionally: treatment time, treatment cost, and / or patient oral health characteristics (e.g., susceptibility to gingivitis).

[0123] exist Figure 3 The processing flow according to this set of specific embodiments is schematically outlined in .

[0124] The algorithm receives information about the type of treatment 38 (e.g., from a dental practitioner's dental record or scheduling system or database). The algorithm also receives an indication of the dental practitioner's treatment style or philosophy (e.g., self-reported in advance or measured in real time). Optionally, the algorithm may also receive information indicating one or more patient oral health characteristics 72.

[0125] The pain prediction algorithm 34 embodied in the processor 32 is configured to predict the expected treatment pain 42 at each step in the procedure, preferably taking into account the dental practitioner's treatment style.

[0126] These estimated pain levels for each treatment step are then visually presented 64 to the dental practitioner for guidance or for personalized training purposes (eg, by displaying a pain prediction score for each step).

[0127] Based on this output 64, the dental practitioner may choose to adjust his or her treatment carefully during portions that are likely to cause discomfort, for example, by speeding up portions that are not predicted to cause much discomfort.

[0128] In particular, more experienced dental practitioners have the skills to perform certain treatments in different styles and methods, for example, faster (and therefore cheaper), but more painful, versus slower and less painful, but more expensive.

[0129] As described above, optionally, the processor 32 is also configured to compare 76 the expected treatment pain (preferably taking into account the dental practitioner's treatment style) to the average treatment pain for similar treatment types across all styles, and feed this information back to the dental practitioner.

[0130] Optionally, the processor can generate and output recommendations 84 to adjust the treatment style to reduce pain, such as: suggesting an alternative treatment style, scheduling an appointment for treatment for more or less time, increasing or adjusting the anesthetic dose, suggesting more or less support staff, and / or providing specific warnings or instructions to the patient in advance.

[0131] Optionally, the output predicted treatment pain may be used to benchmark 82 the treatment styles and skills of different staff within a dental practice or between dental practices, thereby improving the quality of care.

[0132] Expanding on the above example, in some embodiments, means for measuring actual treated pain may additionally be provided (eg, a visual analog scale, or by measuring biosignals such as electrodermal activity).

[0133] exist Figure 4 The extended processing flow according to this group of specific embodiments is schematically outlined in FIG.

[0134] During treatment, the patient's pain level is assessed at each step (or at least a subset) of the treatment 94. Then, after or during the treatment or consultation, the processor 32 is configured to compare 78 the actual pain level 94 with the estimated pain level 42. The comparison can be fed back to the dental practitioner as output information. For example, they can be output to the user through a user interface through a visual presentation 64. This information can be used, for example, as personalized training for the dental practitioner.

[0135] Additionally or alternatively, a comparison 78 between the estimated pain 42 and the actual treatment pain 94 at each step may be used for benchmarking purposes 82. Furthermore, the actual treatment pain 94, along with other features of the treatment, may be used to update a training database 92, which is used to train the prediction algorithm 34. The prediction algorithm may be cyclically retrained using the cyclically updated training database 92.

[0136] This additional feature of comparing the actual pain to the predicted pain at each step can be applied as a more general concept to virtually any embodiment of the present invention.

[0137] Thus, more generally, the method of any embodiment may include the steps of obtaining a metric indicative of an estimated actual treatment pain for at least a subset of treatment steps; for each step in at least a subset of steps, comparing the estimated actual treatment pain to the predicted treatment pain; and generating feedback information for a dental practitioner based on the comparison, e.g., indicating a result of the comparison. Preferably, the metric indicative of an estimated actual treatment pain for each step in the at least a subset of treatment steps may be received in real time during performance of each step in the subset of treatment steps by the dental practitioner. Optionally, the comparison and feedback generation is performed in real time upon receipt of the estimated actual treatment pain for each step. This allows the dentist to adjust their approach for each step in real time.

[0138] With respect to means for obtaining a metric indicative of treatment pain, technical means for doing so have in fact been described earlier in this document, and the reader is referred to these for more complete details. As a brief summary, a metric indicative of estimated actual treatment pain for at least a subset of treatment steps may include: receiving a biosignal of the subject from a biosignal sensing device coupled to the subject during each step of at least a subset of treatment steps; identifying physiological response events in the biosignal; and determining a metric indicative of estimated actual treatment pain for each said treatment step based on the physiological response events identified for that step.

[0139] As a further feature of any embodiment, a device for measuring the treatment style of the dental practitioner in real time (e.g., via pressure and motion sensors integrated into the treatment tool) may optionally be provided. Thus, the prediction of the pain level for each step may optionally be updated in real time based on changes in treatment style.

[0140] In more detail, the method may also include, in some embodiments: receiving data indicating a motion and / or force pattern of a dental tool during at least one step of the treatment; and applying a treatment style analyzer algorithm configured to estimate a treatment style associated therewith based on the motion and / or force pattern. This process may be performed for each step of the treatment. The treatment style analyzer algorithm may be a machine learning algorithm trained to associate a specific pattern of force and / or motion with a specific treatment style, or it may be a rule-based algorithm that applies predefined criteria, such as threshold ranges for different force / motion parameters, to classify treatment styles.

[0141] In some embodiments, the method may further include: re-running the prediction algorithm using the treatment style estimated by the treatment style analyzer algorithm to generate a new prediction of the measure of treatment pain for the at least one step; and generating a user-perceivable output indicative of the new prediction.

[0142] In some embodiments, the pain prediction algorithm, or another algorithm employed by the method and processing unit, may be adapted to have communication access to a database 92 for training the prediction algorithm. The method may also include correlating information about an upcoming scheduled treatment session with similar situations in a training database, for example based on treatment type, treatment style and / or patient oral health characteristics. This may be accomplished by querying the database to find such similar previous situations. For example, a nearest neighbor approach may be employed. Based on similar specific cases identified from the database, the expected treatment pain for different steps of the upcoming treatment may be predicted based on measurements of pain recorded in the identified previous cases. Thus, this differs from the prediction methods discussed so far in that, instead of or in addition to using a trained machine learning algorithm to generate predictions that rely on learning based on an entire population, a simple direct query of the training database itself is used to find similar cases and make pain predictions based thereon.

[0143] As mentioned above, the present invention can be embodied in software. Therefore, another aspect of the present invention is a computer program product comprising code means, which is configured to, when running on a processor, cause the processor to perform a method according to the examples or embodiments of the present invention described in this document or according to any claim of this patent application.

[0144] The embodiments of the present invention described above employ a processing unit. The processing unit may generally include a single processor or multiple processors. It may be located in a single containing device, structure or unit, or it may be distributed between multiple different devices, structures or units. Therefore, references to a processing unit that is adjusted or configured to perform a specific step or task may correspond to a step or task performed by any one or more of a plurality of treatment components, individually or in combination. Those skilled in the art will understand how to implement such a distributed processing device. The processing unit may include a communication module or input / output for receiving data or outputting data to other components.

[0145] The one or more processors of the processing unit can be implemented in software and / or hardware in a variety of ways to perform the various functions required. The processors are typically implemented as one or more microprocessors that can be programmed using software (e.g., microcode) to perform the required functions. The processor can be implemented as a combination of dedicated hardware to perform certain functions and one or more programmed microprocessors and related circuit systems to perform other functions.

[0146] Examples of circuitry that may be employed in various embodiments of the invention include, but are not limited to, conventional microprocessors, application specific integrated circuits (ASICs), and field programmable gate arrays (FPGAs).

[0147] In various implementations, the processor may be associated with one or more storage media such as volatile and non-volatile computer memory (such as RAM, PROM, EPROM, and EEPROM). The storage media may be encoded with one or more programs that, when executed on one or more processors and / or controllers, perform the desired functions. The various storage media may be fixed within the processor or controller, or may be transportable so that one or more programs stored thereon may be loaded into the processor.

[0148] By studying the drawings, the disclosure and the appended claims, those skilled in the art can understand and effect variations to the disclosed embodiments in practicing the claimed invention. In the claims, the word "comprising" does not exclude other elements or steps, and the indefinite article "a" or "an" does not exclude a plurality.

[0149] A single processor or other unit may fulfill the functions of several items recited in the claims.

[0150] The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage.

[0151] The computer program may be stored / distributed on an appropriate medium, such as an optical storage medium or solid-state medium provided together with or as part of other hardware, but may also be distributed in other forms, such as via the Internet or other wired or wireless telecommunication systems.

[0152] If the term "suitable for" is used in the claims or the specification, it should be noted that the term "suitable for" is intended to be equivalent to the term "configured to".

[0153] Any reference signs in the claims should not be construed as limiting the scope.

Claims

1. A computer-implemented method (8) for providing guidance to a dental practitioner regarding dental treatment, the method comprising: Running a prediction algorithm (10), the prediction algorithm comprising at least one trained machine learning algorithm, wherein the prediction algorithm is configured to: Receiving (12) input variables, the input variables comprising at least: a type of treatment required by the patient, the type of treatment being selected from a set of predefined possible treatment types; generating (14) an output prediction data set comprising a prediction measure of treatment pain for each step in a series of steps included in the input treatment type; and Running a guidance algorithm (20), the guidance algorithm is configured to: Accessing (22) the prediction data set; generating (24) guidance information indicative of predicted treatment pain for at least a subset of the series of treatment steps; controlling (26) a user interface to generate at least one user-perceivable output indicative of the guidance information; wherein the prediction algorithm comprises a machine learning algorithm pre-trained using a training database comprising a corresponding data entry for each of a plurality of previous visits to a set of different dental providers for a plurality of different patients, and wherein each visit entry comprises at least an indication of: the type of treatment administered; and A measure indicating the pain of treatment for each step in a series of steps comprised by the treatment.

2. The method of claim 1, wherein the prediction algorithm is configured to receive input variables, the input variables further comprising: A treatment style to be performed by the dental practitioner, the treatment style being selected from a predefined set of possible treatment styles.

3. A method according to claim 1 or 2, comprising controlling the user interface to generate an ordered sequence of user-perceivable outputs, each user-perceivable output indicating the predicted treatment pain for a treatment step in the series of treatment steps.

4. The method according to any one of claims 1 to 3, further comprising: accessing a scheduling database storing records indicating scheduled therapy sessions, each record including at least an indication of a type of therapy to be performed; receiving information indicative of a treatment style to be implemented by said dental practitioner, and optionally wherein said treatment style to be implemented is also included as part of each data record in said scheduling database; Input variables for the prediction algorithm are generated based on the accessing step and the receiving step.

5. A method according to claim 2 or any one of claims 3 to 4 when dependent on claim 2, wherein each visit entry in the training database further includes the following indication: A treatment style, wherein the treatment style is one of the set of predefined treatment styles.

6. The method according to any one of claims 1 to 5, in, within each visit entry, for each of the one or more pain response episodes for each treatment step, the metric indicating treatment pain for each step in the series of steps corresponds to a total measured amount of pain corresponding to the product of the measured pain response level and the time the pain response level was experienced; wherein, within each visit entry, the metric indicative of treatment pain for each step in the series of steps comprises a maximum or peak pain level experienced during the corresponding treatment step; and / or wherein, within each visit entry, the metric indicative of treatment pain for each of the series of steps comprises a metric of a maximum rate of change in pain level experienced during each of the series of treatment steps.

7. The method according to any one of claims 1 to 6, further comprising: obtaining a metric indicative of estimated actual treatment pain for at least a subset of said steps of said treatment; For each subset of at least one subset of said steps, comparing the estimated actual treatment pain to the predicted treatment pain; Feedback information is generated to the dental practitioner based on the comparison, for example indicating a result of the comparison.

8. A method according to claim 7, wherein during each of the subsets of the steps of the treatment performed by the dental practitioner, the metric indicating the estimated actual treatment pain for each of the at least one subset of the steps of the treatment is received in real time, and wherein the comparison and feedback generation is generated in real time upon receiving the estimated actual treatment pain for each step.

9. The method of claim 7 or 8, wherein obtaining a metric indicative of estimated actual treatment pain for at least a subset of the steps of the treatment comprises: receiving, during each of the at least one subset of steps of the treating, a biosignal of the subject from a biosignal sensing device coupled to the subject; identifying a physiological response event in the biosignal; The metric indicative of estimated actual treatment pain for each of the steps of the treatment is determined based on the physiological response events identified for the steps.

10. The method according to any one of claims 1 to 9, further comprising: receiving data indicative of a movement and / or force pattern of a dental tool during at least one step of said treatment; as well as A treatment style analyzer algorithm is applied, the treatment style analyzer algorithm being configured to estimate a treatment style associated therewith based on the motion and / or the force pattern.

11. The method according to claim 10, further comprising: re-running the prediction algorithm using the treatment style estimated by the treatment style analyzer algorithm to generate a new prediction of the measure of treatment pain for the at least one step; as well as A user-perceivable output is generated that is indicative of the new prediction.

12. A computer program product comprising code means configured, when executed on a processor, to cause the processor to perform the method according to any one of claims 1 to 11.

13. A processing unit (32), comprising: Input / output; as well as One or more processors, suitable for: A prediction algorithm (34) is run, wherein the prediction algorithm is configured to: receiving input variables, the input variables comprising at least: a type of treatment desired by a patient (38), the type of treatment being selected from a predefined set of possible treatment types, and generating an output prediction data set (42) comprising a prediction measure of treatment pain for each step in a series of steps included in the type of input treatment; and Running a guidance algorithm (20), the guidance algorithm is configured to: accessing the prediction dataset; generating guidance information (52) indicating predicted treatment pain for at least a subset of the series of treatment steps; controlling a user interface (62) to generate at least one user-perceivable output (64) indicative of the guidance information; wherein the prediction algorithm comprises a machine learning algorithm pre-trained using a training database comprising a corresponding data entry for each of a plurality of previous visits to a set of different dental providers for a plurality of different patients, and wherein each visit entry comprises at least an indication of: the type of treatment administered; and A measure indicating the pain of treatment for each step in a series of steps comprised by the treatment.

14. The processing unit of claim 13, wherein the prediction algorithm is configured to receive input variables, the input variables further comprising: A treatment style performed by the dental practitioner, the treatment style selected from a predefined set of possible treatment styles.

15. A processing unit according to claim 13 or 14, wherein the one or more processors are further adapted to control the user interface to generate an ordered sequence of user-perceivable outputs, each user-perceivable output indicating the predicted treatment pain for a treatment step in the series of treatment steps.

16. The processing unit according to any one of claims 13 to 15, wherein the one or more processors are further adapted to: accessing a scheduling database storing records indicating scheduled therapy sessions, each record including at least an indication of a type of therapy to be performed; receiving information indicative of a treatment style to be implemented by said dental practitioner, and optionally wherein said treatment style to be implemented is also included as part of each data record in said scheduling database; The input variables for the prediction algorithm are generated based on the accessing step and the receiving step.

17. A processing unit according to claim 14 or any one of claims 15 to 16 as dependent on claim 14, wherein each visit entry of the training database further includes the following indication: A treatment style, wherein the treatment style is one of the predefined set of treatment styles.

18. The processing unit according to any one of claims 13 to 17, in, within each visit entry, for each of the one or more pain response episodes for each treatment step, the metric indicating treatment pain for each step in the series of steps corresponds to a total measured amount of pain corresponding to the product of the measured pain response level and the time the pain response level was experienced; wherein, within each visit entry, the metric indicative of treatment pain for each step in the series of steps comprises a maximum or peak pain level experienced during the corresponding treatment step; and / or wherein, within each visit entry, the metric indicative of treatment pain for each of the series of steps comprises a metric of a maximum rate of change in pain level experienced during each of the series of treatment steps.

19. The processing unit according to any one of claims 13 to 18, wherein the one or more processors are further adapted to: obtaining a metric indicative of estimated actual treatment pain for at least a subset of said steps of said treatment; For each subset of at least one subset of said steps, comparing the estimated actual treatment pain to the predicted treatment pain; Feedback information is generated to the dental practitioner based on the comparison, for example indicating a result of the comparison.

20. A processing unit according to claim 19, wherein during each of the subsets of the steps of the treatment performed by the dental practitioner, the metric indicating the estimated actual treatment pain for each of the at least one subset of the steps of the treatment is received in real time, and wherein the comparison and feedback generation is generated in real time upon receiving the estimated actual treatment pain for each step.

21. The processing unit of claim 19 or 20, wherein obtaining a metric indicative of estimated actual treatment pain for at least a subset of the steps of the treatment comprises: receiving, during each of the at least one subset of steps of the treating, a biosignal of the subject from a biosignal sensing device coupled to the subject; identifying a physiological response event in the biosignal; The metric indicative of estimated actual treatment pain for each of the steps of the treatment is determined based on the physiological response events identified for the steps.

22. The processing unit according to any one of claims 13 to 21, wherein the one or more processors are further adapted to: receiving data indicative of movement and / or force patterns of a dental tool during at least one step of said treatment; and A treatment style analyzer algorithm is applied, the treatment style analyzer algorithm being configured to estimate a treatment style associated therewith based on the motion and / or the force pattern.

23. The processing unit according to claim 22, wherein the one or more processors are further adapted to: re-running the prediction algorithm using the treatment style estimated by the treatment style analyzer algorithm to generate a new prediction of the measure of treatment pain for the at least one step; and A user-perceivable output is generated that is indicative of the new prediction.

Images