Personalized Message Scheduling Using ML for User Adherence
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing digital therapeutics face challenges in selecting optimal times for message delivery, leading to sub-optimal engagement, resource wastage, and reduced adherence due to message blindness and reliance on inaccurate user-indicated preferences.
Innovation Solution
A machine learning-based messaging service that adaptively generates personalized schedules for message delivery by clustering user interaction data to identify optimal times, reducing resource consumption and enhancing user engagement and adherence.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If messages are sent at fixed intervals or based on user preferences, then message delivery is simple and consistent, but user engagement decreases and resource waste increases due to sub-optimal timing
Solution Approach 1:
The system dynamically adjusts message delivery schedules based on real-time analysis of user interaction patterns. Instead of fixed intervals, the machine learning model continuously learns and adapts to individual user behavior, determining optimal delivery moments that maximize engagement while minimizing resource consumption from sending messages at ineffective times.
Solution Approach 2:
The system implements a feedback loop where user interactions with messages are continuously monitored and fed back into the machine learning model. This feedback mechanism allows the system to learn from past interactions and refine future message timing decisions, improving engagement rates while avoiding resource waste on messages unlikely to be seen or acted upon.
2Reliability
If messages are delivered frequently to maintain user engagement, then therapeutic efficacy may improve, but message blindness occurs and user adherence deteriorates
Solution Approach 1:
The system applies partial action by selectively delivering messages only when the machine learning model predicts a high probability of user engagement. Rather than sending messages at every possible opportunity, the system judiciously chooses timing and frequency to maintain therapeutic efficacy while avoiding the harmful effect of message blindness caused by excessive or poorly timed communications.
Solution Approach 2:
The system changes the parameter of message delivery timing based on learned user patterns. By adjusting when messages are sent according to individual user behavior parameters (such as typical interaction times, response patterns, and engagement levels), the system maintains therapeutic effectiveness while preventing message blindness that occurs with frequent, uniform messaging.
3Adaptability or versatility
If user preferences for message timing are collected, then message delivery can be personalized, but accuracy is limited due to reliance on self-reporting and inability to adapt to changes
Solution Approach 1:
The system enables self-service personalization by automatically learning user timing preferences through observation of actual interaction behavior, eliminating the need for users to manually report their preferences. The machine learning model continuously adapts to changing user patterns without requiring users to update their self-reported preferences, thereby maintaining high timing accuracy even as user behavior evolves.
Solution Approach 2:
The system replaces the mechanical approach of collecting and relying on user self-reported preferences with an intelligent machine learning system that automatically infers optimal timing from actual behavioral data. This substitution dramatically improves measurement precision by using objective interaction patterns rather than subjective self-reporting, while maintaining adaptability to changing user needs.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Aspects of the present disclosure are directed to systems and methods of dynamically generating individualized or personalized times for providing messages over networked environments. The computing system may obtain, for a user device, an event dataset identifying a plurality of interaction times corresponding to a plurality of interactions over an instant time window by a user with an application on the user device to address a condition of the user. The computing system may apply the event dataset to a machine learning (ML) model. The computing system may generate based on applying the event dataset to the ML model, a defined time at which a message is to be provided to the user device during a subsequent time window The computing system may provide for presentation on the user device to address the condition of the user, the message in accordance with the defined time.