IP Library Granted Patent US 12,658,287
Granted Patent B2
US 12,658,287 · App. 19/017,291 · Granted Jun 16, 2026

Systems and methods for mental health assessment

Inventors: Elizabeth Shriberg (Berkeley, CA); Michael Aratow (Mountain View, CA); Mainul Islam (San Francisco, CA); Amir Hossein Harati Nejad Torbati (Toronto, CA); Tomasz Rutowski (Gdansk, PL); David Lin (Foster City, CA); Yang Lu (Waterloo, CA); Farshid Haque (San Francisco, CA); Robert D. Rogers (Pleasanton, CA)
Assignee: Ellipsis Health, Inc.
G16H10/20A61B5/164A61B5/165A61B5/4803G09B19/00G10L25/66G16H50/20G16H50/30A61B5/4088A61B5/7275G06F16/24G10L15/18
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,658,287
App. No.
19/017,291
Granted
Jun 16, 2026
Kind
B2
Abstract

The present disclosure provides systems and methods for assessing a mental state of a subject in a single session or over multiple different sessions, using for example an automated module to present and/or formulate at least one query based in part on one or more target mental states to be assessed. The query may be configured to elicit at least one response from the subject. The query may be transmitted in an audio, visual, and/or textual format to the subject to elicit the response. Data comprising the response from the subject can be received. The data can be processed using one or more individual, joint, or fused models. One or more assessments of the mental state associated with the subject can be generated for the single session, for each of the multiple different sessions, or upon completion of one or more sessions of the multiple different sessions.

Claims (40)

1 . A system for providing an engaging and interactive conversation with a subject, the system comprising:

one or more computer processors;

a memory comprising machine-executable instructions;

wherein, upon execution, the instructions cause the one or more computer processor to conduct an interactive conversation with the subject using a screening or monitoring system by:

a) obtaining conversation data from the subject during the interactive conversation;

b) assessing at least one state of the subject in real-time during the interactive conversation based on the conversation data using one or more machine learning model, wherein the one or more machine learning model comprises at least one natural language processing (NLP) model, at least one acoustic model, and a composite model that weights and combines outputs from the at least one NLP model and the at least one acoustic model to access the at least one state;

c) selecting or adaptively modifying a message to communicate with the subject based in part on the at least one state;

d) communicating the message to the subject.

2 . The system of claim 1 , wherein at least one of selecting or adaptively modifying the message and communicating the message are adapted based on at least one of environmental data, metadata, demographic data, biometric data, preferences of the subject, and topics of interest to the subject.

3 . The system of claim 2 , wherein adapting the communicating of the message includes at least one of:

adapting a synthetic voice,

adapting physical attributes of an avatar,

adapting language, and

adapting textual data.

4 . The system of claim 1 , wherein assessing the at least one state of the subject identifies one or more significant elements in the conversation data, and selecting or adaptively modifying the message comprises selecting a follow-up message based on the one or more significant elements.

5 . The system of claim 1 , wherein a descriptive analytics module applies interpretable labels to the conversation data from the subject.

6 . The system of claim 1 , wherein a machine learning model estimates truthfulness of contributions of the subject in the conversation data and uses the estimate to adapt the communicating of the message.

7 . The system of claim 1 , wherein the one or more machine learning model assess the at least one state of the subject based in part on past assessments of the at least one state of the subject.

8 . The system of claim 1 , wherein the conversation data comprises at least one of speech data, textual data, and visual data.

9 . The system of claim 1 , wherein the one or more machine learning model further comprises at least one visual model and the composite model combines outputs from the at least one visual model with the outputs from the at least one NLP model and the at least one acoustic model to assess the at least one state.

10 . The system of claim 1 , wherein the screening or monitoring system pauses the communication of the message to the subject for a pause duration to avoid interrupting the obtaining of the conversation data, wherein the pause duration is long enough to avoid interruption of the subject if the subject starts speaking and short enough to maintain engagement.

11 . A method for providing engaging and interactive conversation with a subject, the method comprising:

conducting an interactive conversation with the subject using a screening or monitoring system by:

a) obtaining conversation data from the subject during the interactive conversation;

b) assessing at least one state of the subject in real-time during the interactive conversation based on the conversation data using one or more machine learning model, wherein the one or more machine learning model comprises at least one natural language processing (NLP) model, at least one acoustic model, and a composite model that weights and combines outputs from the at least one NLP model and the at least one acoustic model to assess the at least one state;

c) selecting or adaptively modifying a message to communicate with the subject based in part on the at least one state;

d) communicating the message to the subject.

12 . The method of claim 11 , wherein at least one of selecting or adaptively modifying the message and communicating the message are adapted to based on at least one of environmental data, metadata, demographic data, biometric data, preferences of the subject, and topics of interest to the subject.

13 . The method of claim 12 , wherein adapting the communicating of the message includes at least one of:

adapting a synthetic voice,

adapting physical attributes of an avatar,

adapting language, and

adapting text messages.

14 . The method of claim 11 , wherein assessing the at least one state of the subject identifies one or more significant elements in the conversation data, and selecting or adaptively modifying the message comprises selecting a follow-up message based on the one or more significant elements.

15 . The method of claim 11 , wherein a descriptive analytics module applies interpretable labels to the conversation data from the subject.

16 . The method of claim 11 , further comprising estimating truthfulness of contributions of the subject in the conversation data and wherein communicating the message to the subject is adapted based in part on the estimated truthfulness.

17 . The method of claim 11 , wherein the one or more machine learning model assess the at least one state of the subject based in part on past assessments of the at least one state of the subject.

18 . The method of claim 11 , wherein the conversation data comprises at least one of speech data, textual data, and visual data.

19 . The method of claim 11 , wherein the one or more machine learning model further comprises at least one visual model and the composite model combines outputs from the at least one visual model with the outputs from the at least one NLP model and the at least one acoustic model to assess the at least one state.

20 . The method of claim 11 , wherein the screening or monitoring system pauses the communicating or the message to the subject for a pause duration to avoid interrupting the obtaining of the conversation data, wherein the pause duration is long enough to avoid interruption of the subject if the subject starts speaking and short enough to maintain engagement.

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Jul 6, 2025
From: SHRIBERG, ELIZABETH E.; ARATOW, MICHAEL; ISLAM, MAINUL; TORBATI, AMIR HOSSEIN HARATI NEJAD; RUTOWSKI, TOMASZ; LIN, DAVID; LU, YANG; HAQUE, FARSHID; ROGERS, ROBERT
To: ELLIPSIS HEALTH, INC.
Reel/Frame 071612/0321 →
Continuity (22)
Continuation 18430067 · Feb 1, 2024
Continuation 17671337 · Feb 14, 2022
Continuation 17400066 · Aug 11, 2021
Continuation 17129859 · Dec 21, 2020
Continuation 16918624 · Jul 1, 2020
Continuation 16560720 · Sep 4, 2019
Continuation 16523298 · Jul 26, 2019
Continuation PCTUS2019037953 · Jun 19, 2019
Provisional Application 62755361 · Nov 2, 2018
Provisional Application 62754541 · Nov 1, 2018
Provisional Application 62754534 · Nov 1, 2018
Provisional Application 62754547 · Nov 1, 2018
Provisional Application 62749672 · Oct 24, 2018
Provisional Application 62749663 · Oct 23, 2018
Provisional Application 62749654 · Oct 23, 2018
Provisional Application 62749669 · Oct 23, 2018
Provisional Application 62749113 · Oct 22, 2018
Provisional Application 62733568 · Sep 19, 2018
Provisional Application 62733552 · Sep 19, 2018
Provisional Application 62725356 · Aug 31, 2018
Provisional Application 62687176 · Jun 19, 2018
Related Publication 20250226066A1 · Jul 10, 2025
References Cited (7)
US 20170004260A1 · Moturu · 2017 [cited by examiner]
US 20170061966A1 · Marcheret · 2017 [cited by examiner]
US 20180286429A1 · Bostick · 2018 [cited by examiner]
US 20190042988A1 · Brown · 2019 [cited by examiner]
US 20190213465A1 · Avrahami · 2019 [cited by examiner]
US 20210030308A1 · Grace · 2021 [cited by examiner]
US 20210043208A1 · Luan · 2021 [cited by examiner]