US 12,257,025 B2
AI enabled multisensor connected telehealth system
Marcus Charles Bernard Soori-Arachi, Fort Myers, FL (US)
Assigned to O/D Vision Inc., Sanibel, FL (US)
Filed by O/D Vision Inc., Sanibel, FL (US)
Filed on Sep. 13, 2024, as Appl. No. 18/885,524.
Application 18/183,932 is a division of application No. 29/830,662, filed on Mar. 14, 2022.
Application 18/885,524 is a continuation in part of application No. 18/409,744, filed on Jan. 10, 2024.
Application 18/409,744 is a continuation in part of application No. 18/183,932, filed on Mar. 14, 2023, granted, now 11,877,831, issued on Jan. 23, 2024.
Claims priority of provisional application 63/424,048, filed on Nov. 9, 2022.
Claims priority of provisional application 63/319,738, filed on Mar. 14, 2022.
Prior Publication US 2025/0000361 A1, Jan. 2, 2025
Int. Cl. G06K 9/00 (2022.01); A61B 5/00 (2006.01); G16H 20/00 (2018.01); G16H 50/20 (2018.01)
CPC A61B 5/0022 (2013.01) [A61B 5/746 (2013.01); G16H 20/00 (2018.01); G16H 50/20 (2018.01)] 17 Claims
OG exemplary drawing
 
1. A multisensor-connected, device/edge/cloud AI-enabled, patient-centered health record-storing, telehealth-enhancing system for assisting a provider with differential diagnosis and standard of care and assisting a user/patient with health concern early warning and/or diagnosis, the system comprising:
a multi-sensor medical device comprising:
a plurality of sensors comprising at least seven of: a high-magnification camera module, a motorized camera module, a stethoscope module, an infrared thermopile sensor module, an electrocardiogram (EKG) sensor module, a pulse oximeter module, a body composition monitor module, a glucometer module, a hematology analyzer module, and/or a gyroscope sensor;
a housing enclosing at least the plurality of sensors, a system on chip (SoC) processor, the wireless transceiver, a battery, and the display;
the SoC configured to:
receive patient information from the plurality of sensors;
preprocess the patient information by applying an adaptive feature extraction algorithm;
securely transmit the preprocessed patient information to a cloud-based platform using an end-to-end encryption scheme;
the cloud-based platform comprising:
a secure data storage system for storing the preprocessed patient information;
a computing cluster configured to:
analyze the preprocessed patient information using an ensemble of deep learning models, each model being trained on a specific type of telehealth data and fine-tuned using transfer learning and domain adaptation techniques, the ensemble of deep learning models comprising a modular, extensible architecture and a multi-modal attention fusion module;
generate a ranked list of potential diagnoses and a likelihood score for each potential diagnosis using the ensemble of deep learning models;
an interactive telehealth module configured to communicate with a provider device and the multisensor medical device and transmit information among them and the cloud-based platform, the interactive telehealth module configured to:
send the ranked list of potential diagnoses, the likelihood scores, and a visualization of the factors contributing to each diagnosis to the provider device, the visualization comprising an attention mechanism that highlights a predefined set of salient features for each diagnosis;
receive feedback from the provider device indicating an appropriateness of the potential diagnoses and any additional insights;
send appropriate next steps and resulting insights and documentation to cloud storage, cloud-based AI modules, edge devices of approved users, and the multi-sensor medical device; and
an interactive telehealth portal configured to allow a provider device to communicate with the cloud-based platform, an edge compute node, and the multi-sensor medical device and also displaying data of various forms on a dashboard, dashboard data comprising a live voice and/or video call, a patient chart, raw live sensor data, pre-recorded sensor data, subjective statements of the patient, prior health records, and the above-referenced telehealth module visualization enabling feedback and continuous training.