17H080_Custodio

We develop an AI-based app taking care of patients with mental issues by analyzing their phone activity. The goal is to protect patients from getting intense symptoms and prevent inpatient treatment.

Comment 2

Inspiration

As a result of mental illness, the healthcare system's direct costs amount to around 44.4 billion euros per year in Germany. The total expenses, including indirect costs, are estimated at approximately 147 billion euros per year for Germany. That corresponds to a share of 4.8 % of gross domestic product. The overwhelming share of these costs is reimbursed by a relatively small number of patients (Heavy Users), mainly diagnosed with schizophrenia, bipolar disorders, and personality disorders. With average in-patient cases of more than 30 days and multiple hospital admissions per year, each prevented hospital admission is saving tremendous amounts of money. With an average age between 20 and 30 years, most of them carry around a phone filled with highly diagnostic personal data.

We want to use this data for an AI-based app analyzing the patient's smartphone activity. This solution will take care of users with chronic mental issues and help therapists, psychiatrists, and the patient himself to take preemptive actions.

The goal is to protect patients from getting intense symptoms and prevent residential treatment. 

What it does

We identified three stakeholders as the primary profiteers of a Custodio introduction: Patients, Physicians and Health insurance companies

Patients face problems as:

  • Huge amounts of medications cause strong side effects
  • Social and professional life suffers because of long periods of sick leaves
  • Lower life expectancy caused by disease and medication

Custudio improvement:

  • Improved prevention leads to less medication thus fewer side effects
  • Increased life quality thanks to more planning certainty and less illness
  • Increased life expectancy thanks to a healthier lifestyle and less exposure for liver and kidney

Physicians face problems as:

  • Psychotherapeutic treatment is less effective as the disease is already progressed
  • Patients abandon treatment as they suffer from medication side effects
  • High referral of patients in inpatient stay

Custudio improvement:

  • Early-stage treatment with psychotherapy is more effective
  • Physicians have a closer treatment success as their patients have to handle with fewer side-effects and symptoms
  • Fewer patients need to be referred to psychiatry

Health insurance company face problems as:

  • “Heavy users” causing the principal share in mental health costs
  • Long hospital stays are costly
  • A lot of psychotropic prescriptions and additional health costs of long-term users

Custudio improvement:

  • Closer monitoring of “heavy users” and prevention of relapse and in-patient hospital stays
  • The improved monitoring may allow for shorter in-patient stays thanks to a more efficient ambulatory care
  • Fewer psychotropic drug prescriptions and less long-term damages caused by medication

How we will build it

We want to use the patient's smartphone data for an AI-based app analyzing the patient's smartphone activity.

The prediction is based on:

  • App Usage
  • Activity Tracking
  • Social Media Texts

Challenges we ran into

Our main challenge is getting access to data. Mental Health is a highly sensitive topic, and most are not comfortable sharing data. Social media apps that focus on emotional and mental health support - such as VENT or TalkLife - could be used to build a language model.

However, to build a classifier, the texts will still need to be associated with more severe or less severe disease phases.

Accomplishments that we're proud of

  • We are a team of people bringing together all the necessary skills to meet this challenge.
  • We developed a sophisticated concept improving the situation of all stakeholders
  • We draw up a business plan identifying financial needs, financials outcomes, growth potential, distribution channels and partners

What we learned

Multiple research and data are already available for our prototype development. To expand our target group and build a more robust detection system, we should conduct a study

What's next for 17H_Custodio

The application will require multiple data sources from affected patients. It will thus require data acquisition in the form of a study on heavy users.

We are looking for the following advisors and partners to help us to realize our solution:

  • Universities/ Healthcare Institutions/ Scientists/ Physicians helping us to collect data
  • Representatives of the Bundesinstitut für Arzneimittel und Medizin Produkte (Experts on DIGA)
  • Venture Capital Fonds/Business Angels
  • Representatives of Health Insurance Companies
Share this project:

Updates