Big Data in Pe­diatric On­co­lo­gy and Rare Di­sea­ses

The future of child­hood cancer tre­at­ment: a smart, me­di­cal­ly safe home 
The aim of this Sandpit project is to enable proac­tive, per­so­na­li­zed tre­at­ment of child­ren with cancer in their home en­vi­ron­ment. “Looking after child­ren with cancer is emo­tio­nal­ly taxing,” says project spo­kes­per­son Dominik Schön­dorf from Uni­ver­si­ty Hos­pi­tal Saar­land. “Parents are often unsure when they require medical as­si­s­tan­ce, and es­pe­ci­al­ly when they need to take their child to hos­pi­tal outside of sche­du­led ap­point­ments. Our project is in­ten­ded to prevent un­ne­cessa­ry stays in hos­pi­tal.” The Sandpit project aims to put in place in­fra­st­ruc­tu­re that will make use of big data, ar­ti­fi­ci­al in­tel­li­gence, and machine lear­ning to predict the medical needs of child­ren before an emer­gen­cy si­tua­ti­on arises. The idea is that the patient’s home en­vi­ron­ment will provide a con­stant stream of health in­for­ma­ti­on so that warning signs can be spotted at an early stage. These could be changes in vital signs, altered be­ha­vi­or, or en­vi­ron­men­tal factors. It is hoped that in­tel­li­gent net­wor­king between home and hos­pi­tal will prio­ri­ti­ze quality of care for pa­ti­ents and quality of life for fa­mi­lies. 

Buil­ding SHARE: the Syn­the­tic Health Data Re­po­sito­ry
An in­ter­na­tio­nal team is plan­ning to build a syn­the­tic health data re­po­sito­ry for rare di­sea­ses. The aim is to make stan­dar­di­zed data freely avail­ab­le to re­se­ar­chers to train AI models and to conduct data-driven re­se­arch, fa­ci­li­ta­ting re­se­arch in the field of health­ca­re and phar­maceu­ti­cals. “The need for large volumes of data for AI models in health­ca­re is in­crea­sing ex­po­nen­ti­al­ly,” says project spo­kes­per­son Jannik Schaaf from Goethe Uni­ver­si­ty Frank­furt. “Par­ti­cu­lar­ly for rare di­sea­ses, of which there are more than 7,000, there are hardly any sta­tis­ti­cal­ly re­le­vant patient da­ta­sets and only a few AI models.” Syn­the­tic data is ge­ne­ra­ted ar­ti­fi­ci­al­ly and can reflect sta­tis­ti­cal pro­per­ties of real data. This type of data is vital for cli­ni­cal re­se­arch, es­pe­ci­al­ly in the area of rare di­sea­ses. Another problem the project aims to address con­cerns global com­pa­ra­bi­li­ty because it is often not pos­si­ble to access the data un­der­ly­ing exis­ting trained AI models. SHARE is de­si­gned to close these gaps and provide stan­dar­di­zed data for re­se­ar­chers world­wi­de.