Birkeland Current's Sovrin IoT system used to predict acute health incidents up to 2 weeks in advance

Birkeland Current’s Sovrinti company ended 2023 with the completion of a National Institute on Aging sponsored program whose objective was to predict acute health events in older adults up to 2 weeks in advance of the occurrence. As part of the NIA sponsored and John’s Hopkins led Artificial Intelligence Technology Collaboratory, Sovrinti successfully developed a deep learning algorithm that was able to predict acute health incidents from the previously validated Sovrin IoT Activity of Daily Monitoring (ADL) system. Using home sensor data from Birkeland Current’s 2020 - 2023 ADL research, the developed deep learning program was trained to recognize changes in individual patterns and predict rising risks of acute issues. A full report is available at www.sovrinti.com. A summary is available below.

Birkeland Current Presents NIA Study Results at Gerontological Society 2023 meeting

Birkeland Current presented preliminary results at the 2023 Annual Scientific Meeting of the Gerontological Society of America.

Abstract: This paper presents preliminary correlation analysis results between the Birkeland Current Sovrin IoT in-home sensing system and monthly ADCS-ADL survey data for 117 CareRecipient/CareGiver (CR/CG) pairs as part of a Phase II NIA SBIR study (R44AG065118) conducted from 2021 to 2023. The analysis provides correlation analysis between overall BC ADL Scores and ADCS/ADL Scores as well as comparisons to two-factor IADL/BADL scores and to the four-factor Kahle-Wrobleski ADL scores (W1-W4). The CR population was screened to represent mild to mid-level cognitive decline utilizing the MMSE tool with the expectation that most of the change in ADL scores would be based on IADL groupings. Utilizing percent change from monthly ADL survey scores and associated subscores, the BC system demonstrates significant correlation with the ADCS-ADL score at the level of sensitivity of the ADCS survey and subfactor results. For modest change, the BC system shows significant sensitivity to two factor IADL change (10-20%) with the two-factor BADL sensitivity dominating for 30-40% change. The BC system shows little or no sensitivity to the W3 (communications and engagement) and W4 (outside activities) scores as these are items not directly measurable by the BC system. The correlation with the W2 (Household Activities) provides the highest subfactor correlation at low change levels and most closely represents the BC system approach which is dominated by household activity measures. Utilizing the correlation approach, the BC system is able to demonstrate sensitivity to a single point change in the two factor BADL score and a one-point change in the two factor IADL score which matches the consensus sensitivity of the ADL survey. All data presented is statistically significant at an alpha value of 0.05.