Patent Application
20240105300
It is common for physicians to prescribe medications for their patients. After a prescription is generated for a patient by a physician, the patient or the physician conveys the prescription to a pharmacy.
The pharmacy “fills” the prescription by preparing a container that contains the medicine specified by the prescription, in a form, dosage, and quantity specified by the prescription, and delivers it to the patient, such as in person or by mail.
Some prescriptions are written in such a way as to provide for one or more “refills,” such that, after the patient takes all of the medicine in the original container, the patient can obtain one or more similar containers in order to continue taking the medicine. In order to do so, the patient contacts the physician or the pharmacy, such as by going to one of these places in person, or placing a phone call and speaking to a person who works there.
The inventors have recognized significant disadvantages in conventional approaches available to patients to request prescription refills. In particular, talking to a person at the pharmacy or the physician's office can be difficult for a patient in many respects: (1) it can take time and effort to travel to a physical location, or determine the correct phone number to call; (2) there are many hours of the day and night when no one is working to speak to the patient, and additional hours when everyone who is working is occupied, requiring the patient to wait to speak to someone, or come back or call back later; (3) it generally takes time for the person to whom the patient speaks to access electronic records about the prescription, which they need in order to determine whether a refill can be ordered and to order it, and this access is subject to manual error; (4) in some cases, the patient may have trouble understanding the speech of the person they speak to, or being understood by them, either in being able to recognize which words are spoken, or in understanding the larger ideas that the speaker is trying to communicate; and (5) conducting a spoken conversation with another person tends to require certain levels of mental, physical, and emotional energy, and each spoken conversation a person conducts can reduce some or all of these levels, making it more difficult to conduct later ones—this can be true both of patients and of workers at pharmacies and physicians' offices.
In response to recognizing these disadvantages, the inventors have conceived and reduced to practice a software and/or hardware facility for receiving prescription refill requests via voice and/or free-text chat conversations between a patient and an automated agent (“the facility”).
The facility conducts a conversation between the patient and an automated agent. In some embodiments, the patient speaks to the automated agent, and the facility generates simulated speech from the agent to the patent. In some embodiments, the patient and the agent send free-form text strings to one another. In some embodiments, these modes are combined, or the conversation is conducted using other similar modalities.
In various embodiments, the patient uses various mechanisms to engage in the conversation, including, for example, a specialized medical assistance application, a voice and/or text chat application, a telephony application, a browser, a plugin to a general-purpose assistant, etc. These can be on a smartphone, a wearable device, a computer system, a POTS handset, a kiosk, or a variety of other devices.
In some embodiments, in the conversation, the facility invites the patient to share their intent for the conversation. Such intents can extend to a broad array of subjects beyond refilling prescriptions. As the conversation proceeds, the facility continues to seek this intent, as well as named entities (“entities”) relevant to the expressed intent, providing context intended to assist the patient in providing this information. For example, when the facility discerns the intent of refilling a prescription, in some embodiments it (1) determines that an entity identifying the particular prescription to be refilled is necessary for fulfilling this intent, and (2) lists the patient's prescriptions to assist the patient in specifying the entity corresponding to the prescription to be refilled. In some embodiments, the facility applies machine learning models to the patient's input in order to discern intent and relevant entities.
Once the facility discerns the intent to refill a prescription and the identification of an entity identifying the prescription to refill, it accesses a record for this prescription, such as one stored for the patient by electronic medical record (“EMR,” or “EHR”) software. Based on the contents of this prescription record, in some embodiments, the facility takes a different path to order the refill. Where the prescription has not expired and a number of refills specified by the physician in writing the prescription has not been reached, the facility proceeds to order the refill from the pharmacy, such as by calling a programmatic ordering interface exposed by the pharmacy; calling an ordering voice-response telephone line operated by the pharmacy with simulated touch-tones and/or computer-generated or manually-recorded voice; calling a live ordering telephone line with computer-generated or manually-recorded voice; or directing a person such as a physician's assistant to call such a live ordering telephone line. Where the prescription has expired or its number of refills has been exhausted, the facility refers a renewal request to the physician who wrote the prescription, or a delegee. If the physician or delegee approves the renewal, the facility proceeds with the order as discussed above. If a problem arises in this process, the facility advises the patient to contact the physician directly.
By operating in some or all of the ways described above, the facility enables the patient to quickly and easily order a prescription refill, using modalities and tools that they find comfortable and effective, at any time that is convenient to them. Also, the facility circumvents many of the categories of errors that can occur in more manual conventional approaches.
Additionally, the facility improves the functioning of computer or other hardware, such as by reducing the dynamic display area, processing, storage, and/or data transmission resources needed to perform a certain task, thereby enabling the task to be permitted by less capable, capacious, and/or expensive hardware devices, and/or be performed with lesser latency, and/or preserving more of the conserved resources for use in performing other tasks. For example, by eliminating the need to order refills by phone, the facility relieves the load imposed on telephony hardware, such as handsets, switches, etc., freeing them up for increased or improved use for other purposes, or permitting them to be replaced with lower-capacity, cheaper alternatives, or even eliminated. Similarly, eliminating the need to order refills in person, the facility relieves the load imposed on vehicles, such as cars, motorcycles, bicycles, buses, taxis, etc., freeing them up for increased or improved use for other purposes, or permitting them to be replaced with lower-capacity, cheaper alternatives, or even eliminated.
In act 304, the facility receives user input, typically responsive to its last conversational interaction from the smart assistant. In act 305, the facility automatically transcribes user input received in the voice modality to text, if this is necessary for the processing of the voice input by the machine learning models used by the facility, and/or in order to be able to display the transcribed text as part of the visual representation of the conversation. In various embodiments, the facility uses various mechanisms to perform the transcription of act 305, including a variety of transcription engines and/or natural language voice models.
Table 2 below shows a sample first user input received in response to the conversational interaction shown above in Table 1.
In act 306, the facility applies one or more machine learning models to extract from the user input—in some cases as transcribed in act 304—intents and entities that it contains. In various embodiments, the facility uses a variety of machine learning model types. In some embodiments, the facility uses a transformer-based machine learning model, such as those described in any of the following, each of which is hereby incorporated by reference in its entirety: “BERT” (available at huggingface.co/docs/transformers/model_doc/bert); J. Devlin, M. W. Chang, K. Lee, K. Toutanova, “BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding,” arxiv: 1810.04805 (available at arxiv.org/abs/1810.04805); B. Portelli, “DiLBERT (Disease Language BERT)” (available at huggingface.co/beatrice-portelli/DiLBERT); W. Siblini, M. Challal, C. Pasqual, “Delaying Interaction Layers in Transformer-based Encoders for Efficient Open Domain Question Answering,” arxiv: 2010.08422 (available at arxiv.org/abs/2010.08422); and K. Roitero, B. Portelli, M. H. Popescu and V. D. Mea, “DiLBERT: Cheap Embeddings for Disease Related Medical NLP,” in IEEE Access, vol. 9, pp. 159714-159723, 2021, doi: 10.1109/ACCESS.2021.3131386. (available at ieeexplore.ieee.org/document/9628010). In cases where the present application conflicts with the document incorporated by reference, the present application controls.
In the example conversation shown in Tables 1 and 2, the facility extracts the intent of refilling a prescription. In act 307, if adequate intents and entities have been extracted to be able to process the prescription refill request, then the facility continues in act 308, else the facility continues in act 302 to exchange another pair of interactions.
Continuing the example, in the second iteration of act 303, the facility formulates the additional interaction shown below, in which it confirms the intent to refill a prescription and assists the user in identifying the particular prescription to refill by listing those that are active for the user in the user's EMR record. In some embodiments, the facility calls an API exposed by the EMR in order to retrieve information about active prescriptions. For example, by using the EPIC EMR, the facility calls its GetPrescriptionInfo (2018) API for this purpose.
Table 3 further shows that, in the second iteration of act 304, the facility receives user input selecting a particular one of these prescriptions, “Metformin,” for refilling. The facility further confirms the selection, and indicates that it will place the refill request order, as discussed below.
In act 308, the facility processes a prescription refill request for the prescription corresponding to the entity extracted from user input. Additional details of act 308 are discussed below in connection with
Those skilled in the art will appreciate that the acts shown in
The various embodiments described above can be combined to provide further embodiments. All of the U.S. patents, U.S. patent application publications, U.S. patent applications, foreign patents, foreign patent applications and non-patent publications referred to in this specification and/or listed in the Application Data Sheet are incorporated herein by reference, in their entirety. Aspects of the embodiments can be modified, if necessary to employ concepts of the various patents, applications and publications to provide yet further embodiments.
These and other changes can be made to the embodiments in light of the above-detailed description. In general, in the following claims, the terms used should not be construed to limit the claims to the specific embodiments disclosed in the specification and the claims, but should be construed to include all possible embodiments along with the full scope of equivalents to which such claims are entitled. Accordingly, the claims are not limited by the disclosure.
