The advent of targeted biologic and synthetic disease modifying antirheumatic drugs (DMARDs) along with early and treat-to-target treatment strategies has dramatically improved patient care in rheumatology. However, e.g. in rheumatoid arthritis only 30% of the patients reach full and sustained remission. In general, targets in immune-mediated diseases (such as remission) are not clearly defined and not very sensitive. In current guidelines, clinical improvement and targets should be achieved after three and six months, respectively. At this point, currently available risk factors and biomarker still do not help to navigate treatment in our patients und thus their role in precision medicine (in rheumatology) at this point remains limited. Recent studies have shown proof of concept that machine learning applications can predict outcome in arthritis. Potentially deep learning or other forms of machine learning can be used to navigate therapy and to avoid over- or undertreatment. Apart from genomic or transcriptomic data, digital biomarker (e.g. by image recognition) and patient reported outcome can improve digital disease prediction as well as other machine learned features such as disease stratification and outlier detection.