Invented by Maddala; Tara, Rothney; Megan, Carlson; Jason, Nistala; Akhil, Waugh; Miles, Civello; Daniel, Geis; Jennifer

Helping people with their mental health is more important than ever. A new patent application outlines a way to use artificial intelligence (AI) to create personal treatment plans for mental health. In this article, we’ll break down what makes this invention special, why it’s needed, how it works, and how it stands out from what came before. Our goal is to make this complex topic easy to understand, whether you’re a parent, student, teacher, or health professional.

Background and Market Context

In recent years, the need for better mental health support has grown quickly. Many young people, especially students, are struggling. Studies show that about 60% of U.S. college students have some kind of mental health problem. These issues include depression, anxiety, eating disorders, and problems with substance use. The COVID-19 pandemic made things even harder, creating more stress and feelings of loneliness in people of all ages.

But there is a big problem: even though many students and young people struggle, only about 40% actually get help. There are many reasons for this. Some are worried about what others will think. Some don’t know where to get help. Others think resources are hard to find or not made for them. Schools and workplaces are trying to do more, but they don’t always have the staff or money to help everyone.

At the same time, technology is a big part of our lives. Most people carry smartphones. Many wear smartwatches or fitness trackers. We use social media, send texts, make calls, and even use apps to help with stress or sleep. These devices collect lots of data—like how much we move, where we go, how we sleep, and how we interact with others. This data can give clues about how someone is feeling or coping.

Companies and researchers started to wonder: What if we could use all this digital data to spot when someone might need help? What if we could use it to suggest the right support, at the right time, for each person? That’s where the idea behind this patent comes in. It aims to make mental health support smarter, faster, and more personal by using AI and data that people already generate every day.

Scientific Rationale and Prior Art

Before this patent, mental health care mostly relied on people reaching out for help or filling out surveys. Doctors and therapists would listen, ask questions, and use their training to suggest what might help. Sometimes, apps would offer self-guided activities, like breathing exercises, mood tracking, or journaling. Some apps could track steps, sleep, or phone use, and some even let people rate their mood each day.

Researchers saw that certain patterns in digital behavior could be linked to mental health. For example, changes in sleep or movement, spending more time at home, or a sudden drop in phone use could signal problems. Also, the way people talk or write—like using more negative words or speaking less—can be a sign of feeling down or anxious.

Some earlier inventions tried to use this kind of data. They might look at just one type of information, like how many steps someone takes, or how often they use social media. Others used simple rules, like sending a reminder if someone hasn’t opened an app in a few days. More advanced ideas used machine learning models to spot patterns. For example, a model might learn that people who start moving less and texting less often are more likely to feel sad.

But there were some big limits:

• Most systems only used one or two kinds of data (like just sleep or just steps).
• They often kept the actual words or messages people sent, raising privacy worries.
• They didn’t always handle missing data well. If someone skipped a check-in, the system might not know what to do.
• They rarely suggested a complete, personal treatment plan. Usually, they just flagged a risk or gave a simple tip.
• They didn’t combine digital data with personal preferences, history, or engagement to make better suggestions.

This new patent builds on what came before but tries to solve these problems. It uses many types of data together, protects privacy by focusing on feelings (not the exact words), and uses powerful AI models to create a full, personal treatment plan for each user.

Invention Description and Key Innovations

The patent describes a system and method that collects data from a user’s device (like a phone or smartwatch) and uses AI to create a treatment plan for mental health. Here’s how it works, in simple steps:

1. Collecting Rich, Multimodal Data

The system gathers many types of data from the user’s devices and apps. This can include:

• Voice data – how someone’s voice sounds, not just what they say (tone, pitch, speed, etc.)
• Text data – not the full messages, but the feelings in the writing (using emojis, punctuation, etc.)
• Location data – places visited, time spent at home, school, or other locations
• App usage – which apps are used, for how long, at what times
• Biometric data – heart rate, sleep patterns, activity, steps, and more
• Self-reported data – how the user says they feel, or answers to short questions
• External data – things like weather, news, or local events that might affect mood

2. Protecting Privacy by Focusing on Sentiment

One of the special things about this invention is how it handles sensitive data. Instead of keeping the actual words from texts or recordings, it uses AI to “encode” the sentiment—the feelings or mood—while throwing away the exact messages. For example, if someone texts “I feel overwhelmed,” the system just notes the feeling (“stressed”), not the exact words. This helps protect privacy.

3. Handling Missing or Incomplete Data

People don’t always answer every survey or share every kind of data. The system’s AI models are trained to work even when some data is missing. If someone doesn’t record a voice prompt one week, or if their smartwatch battery dies, the system can still make a good guess about how they’re doing using the other information it has.

4. Two AI Models Working Together

The system uses two main neural networks (a kind of AI brain):

• The first model looks at all the data and finds patterns in the user’s feelings and behaviors. It creates a “marker” that represents the user’s current mental state and predicts how they might respond to different kinds of help.
• The second model takes this marker and combines it with the user’s profile (like their preferences, age, gender, and how they use the app). It then creates a treatment plan that is tailored just for that person.

The treatment plan could include many kinds of support: suggestions for therapy, group support, changes in sleep habits, medication (if a doctor is involved), or simple exercises to try.

5. Personalization and Ongoing Updates

The invention doesn’t just give one-size-fits-all advice. It learns what works best for each person over time. If someone responds well to a peer support group, the system will remember that. If a user’s habits change—like sleeping less or exercising more—the system updates their plan. It can also notice when things get worse and suggest new kinds of help early, before a crisis happens.

6. Advanced Machine Learning Techniques

The AI can use many different types of machine learning, like principal component analysis (to find key patterns), neural networks (to handle complex data), and time series models (to track changes over days or months). This means the system can spot trends, not just snapshots in time, and can tell if someone’s mental health is getting better or worse.

7. Broad Flexibility and Scalability

This system can work for many groups—students, employees, military members, or anyone who wants to track their mental health. It can be used on phones, computers, or even in the cloud, and can handle large groups as easily as individuals. It can also be set up to work in different settings, from schools to companies to clinics.

8. Easy Integration and Actionable Output

The output is not just a number or alert. The system can give clear, actionable plans and resources. For example, it might recommend a local counselor, a meditation app, or send reminders to drink water or take a walk. It can refer users to the right kind of help, matching their background, interests, and needs. It can also check if users follow the advice and update plans if needed.

9. Ethical Data Use and User Control

The invention is designed to be ethical. Users can choose what data to share. All data can be encrypted and anonymized. If users want to delete their data or change what the system tracks, they can do that.

10. Real-World Testing and Results

The patent describes real-world pilot studies with students, showing that the system can collect lots of useful data and spot early signs of problems. For example, students who walked less and used their phones differently often had higher levels of anxiety or depression. The system could use these clues to reach out and offer help sooner.

Conclusion

This new patent represents a big step forward in making mental health care smarter and more personal. By bringing together many types of data, protecting privacy, and using advanced AI, it can spot problems early and offer the right support to each person. It works even when not all data is available, keeps learning over time, and helps people get the help they need—in a way that fits their life and values.

This technology is more than just a tool. It’s a bridge between the digital world and real human care. As mental health challenges grow, inventions like this offer hope for early detection, better support, and healthier lives for everyone.

Click here https://ppubs.uspto.gov/pubwebapp/ and search 20250218568.